AF101 tests up

88 replies [Last post]
Piotr
Offline
Joined: Jun 9 2006

http://thebrownings.name/WHP034/

addendum 66

Alan,
What do you think BBC position will be on this cam?

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

Yep, it's there now.

I suspect the official position will be that it's an HD camera, but not top grade. So, it'll get approval for use in film-style shoots but not as the main camera. I guess it's footage will be classed in the 'non-HD' quota, including archive and SD footage. But, I don't make the rules or the decisions, I only measure and report (and get paid for so doing).

I'll be posting a revision to the PMW350 as well soon, and the PMW500.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

infocus
Offline
Joined: Jul 18 2003

I note in the report you say:

Quote:
The horizontal and vertical aliases have a centre frequency of 1.224 times the system frequencies. Thisim plies that there is an underlying resolution 1.224 times that of 1920x1080, or 2350x1322, and this furtherimplies that the sensor pixel dimensions are double that or 4700x2644

This is very close to the results I was previously shown, but if the alias circle centres are at 2350x1322, then are these figures not TWICE the underlying resolution? (The alias circle centres representing 2xNyquist?)

So the underlying resolution before aliasing is therefore 1175x661?

Piotr
Offline
Joined: Jun 9 2006

The paper claims (?) that AF100 and 4/3 imager is 1.5 stops away from the 1/2 inch imager DOF-wise.

"If this camera is to be used for HDTV shooting, then it should be clearly understood that it’s only advantage over smaller-format cameras is the smaller depth of field. However, to achieve a smaller depth of field in this camera, relative to, say, a ½” camera, then the lens must be opened by at least 1.5 stops; using an F/2.8 lens on this camera, wide open, will give the same depth of field as on a ½” camera with a lens opened toF/1.6. This camera will not necessarily always deliver short depth of field, large aperture lenses must be used to achieve that".

I'd think it's more like 2.5-3 stops depending on the actual 1/2 inch model imager size , so you'd need to open it up to almost f1.0 to be comparable in DOF to f2.8 on 4/3. Unattainable on a prism/3x sensor construction. And for 4/3 a f0.95 glass is available (Noctilux, Nokton). Matchable by impossible f0.4 glass on 1/2 inch.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

Piotr, unless you've actually measured it, as I have, how can you be so sure I'm wrong?

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Piotr
Offline
Joined: Jun 9 2006

DOF is purely conventional a notion, so I cannot say you are wrong. I can only ask what method od appraising DOF you assume.
I go by the circle of confusion concept (CoC). Whatever CoC value (diameter) we assume for the sharpness limit (in 35mm cinematography it is 1/1000 inch or 0.025mm on the film frame*), for a given visual acuity and viewing conditions only the size of the CoC will describe DOF. Given the Af101 sensor width of 17.8mm and traditional 1/2 inch imager width of 6.4mm (some are wider I understand), the magnification ratio is 2.78 between the two formats. ie. 1/2 inch must be magnified 2.78x more than 4/3 to achieve the same display size. So does the assumed CoC. For the same angle of view and focussing distance the DOF on the two formats should be governed by this 2.78 stop (f number) ratio. N=f/D where N is f number (f stop), "f" focal length, "D" entrance pupil diameter.
I see only linear/proportional relations here, eg. between the CoC diameter and entrance pupil diameter. So from f2.8 on AF100 we should go to ca f1.0 on 1/2 inch (6.4 mm imager width) to match the two DOFs--2.78 or almost 3 stops. Again, for the same AOV and focussing distance.
That was my train of thought, but optics is not really my area of expertise and I can be mistaken of course, will be glad to stand corrected.

*as per American Cinematographer Manual, although a newer more stringent 1/2000 inch standard is also mentioned. This actual value is without influence on the discussed issue. Although it's worth noticing that S35 camera gate is 24.9 mm wide and Af100 sensor 17.8mm wide, so the magnification ratio here is 1.4x. Accordingly the AF100 CoC for cinema should be 0.025mm/1.4=0.018mm to match the cinematic DOF standard.

SimonMW
Offline
Joined: Nov 16 2004

All in all though a pretty damning conclusion as to the cameras performance. The results of the actual clean resolution produced by the camera were pretty close to what a few of us reckoned based upon another chart. One that I might add was poo pooed by a certain well known Panasonic rep who told us that in their tests the camera easily produced 800 lines.

Ho hum.

MAGLINK
Offline
Joined: Mar 8 2007
SimonMW wrote:
All in all though a pretty damning conclusion as to the cameras performance. The results of the actual clean resolution produced by the camera were pretty close to what a few of us reckoned based upon another chart. One that I might add was poo pooed by a certain well known Panasonic rep who told us that in their tests the camera easily produced 800 lines.

Ho hum.

Don't worry about it Simon as a panasonic customer I was well and truly told I was talking bolex when I questioned if they had actually put a new sensor on the 371, it all went very quiet after they had to release a firmware version to remove the filtering that they had slapped on and it gave you a 301 again.

it is amusing the number of excuses now coming out and saying how we now don't need full 1080 resolution for HD, two steps forward and five back as the shallow DOF DSLR brigade seem to get all the attention as they whine the most about their art! :D

infocus
Offline
Joined: Jul 18 2003
Alan Roberts wrote:
Piotr, unless you've actually measured it, as I have, how can you be so sure I'm wrong?

I haven't measured the difference for 1/2" v 4/3, but I have formally measured it to compare 1/3" cameras v 2/3", and the difference in that case is unquestionably 2 stops. For dof comparison, if the sensor area is doubled, the lens must be closed down a stop for the same dof. (Assuming angle of view etc etc is constant.)

Going to an excellent textbook, I found the following:

Quote:
It's all very interesting to be able to calculate actual depth of field, but it's rather more useful to have a comparison between formats, a means of getting the same depth of field in different formats. There's really no need for a diagram or a table, because, for a given lens angle of view, this releationship holds true:

F1/d1 = F2/d2, and so F2=F1*d2/d1

...where the F numbers are lens stop settings, and the d numbers are physical iamge dimensions

Now the same book gives 7.2mm as the width of a 1/2" sensor in 16:9 mode, and a four thirds sensor may be taken to be about 17.8. Putting these numbers into the above formula, then f2.8 on an AF101 equates to about 2.8*7.2/17.8, or about f1.1. In other words, a 2.5 stop difference, not 1.5.

(I think Piotr was using the 4:3 value for the width of the 1/2" sensor - 6.6mm - not the 16:9 figure. Otherwise his reasoning seems correct.)

I'd like to ask again about the fundamental resolution. Your report talks of "an underlying resolution 1.224 times that of 1920x1080, or 2350x1322". Surely that should be 1.224x0.5x1920x1080, or 1175x661?

Surely the centres of the circles of the aliases represent twice the underlying resolution - not the underlying resolution itself?

Hence, the camera is then UPSCALING the fundamental 1175x661 to make 1920x1080 - not downscaling 2350x1322? That would explain the measured results far better.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

No, it's down-scaling. My numbers are right.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

infocus
Offline
Joined: Jul 18 2003
Alan Roberts wrote:
No, it's down-scaling. My numbers are right.

I don’t see how that makes engineering sense. It means effort (in terms of processing and power) to read the sensor on a 2350x1322 basis (2x2 photosite blocks) – which then all goes to waste, to produce a mediocre output which doesn’t even come up to 720p standards. A fundamental 2350x1322 resolution should give a decent 1920x1080 output – but it doesn’t. It would point not just to a poor downconvertor, but a truly abysmal one. I don’t believe it.

Whereas, if the photosites were dealt with in 4x4 blocks, 16 at a time, it would be far, far easier to implement technically with much lower processing and power requirements than the above method. And give exactly the results you have measured – equal red, green, blue resolutions, each of a quarter the total sensor dimensions, and each of a resolution 1175x661.

Why put a lot of effort into getting a result, when the same end result could be obtained far more simply in another way?

And a simpler reading of the sensor (4x4 blocks, instead of 2x2) would likely mean being able to read the sensor faster per frame, hence less rolling shutter - exactly as observed.

Piotr
Offline
Joined: Jun 9 2006
SimonMW wrote:
All in all though a pretty damning conclusion as to the cameras performance. The results of the actual clean resolution produced by the camera were pretty close to what a few of us reckoned based upon another chart. One that I might add was poo pooed by a certain well known Panasonic rep who told us that in their tests the camera easily produced 800 lines.

Ho hum.

I am not as much turned off by the actual resolution figures (or other performance quirks) of the Af100, as by Panasonic's resilience to come clean on this cam. For any professional broadcast/film use we need to know everything possible about our tools, warts and all. Then we simply accept and adapt, or take another tool. Panasonic's behaviour clearly suggests that this camera is not addressed to this group of professionals. To me this stance means I walk away from the Af100 as a viable film and TV production camera. I have no time for such sillyness, inspite of the effort I devoted to test the cam and play with it. And the money spent. Perhaps we'll use it as a c-cam in some productions.
It's a pity, because I think the cam is not that bad at all. I like the picture it delivers, especially with f2 Zuiko Digital zooms. The resolution of 650-680TVL would not be that catastrophal in drama applications, especially drama productions watched on home displays from more than one screen diagonal distance. In fact the average cinema delivers ca 800 lines of resolution in projection and nobody complains, even those sitting closer than one screen diagonal. Nobody comes out of cinema talking about the picture resolution.
But to put a camera on a set I need to believe in it. Like in our SDX900 or their replacements now the HPX3100. Fabulous cams. AF100 will remain in the hands of "dvx users"and "dv infoed", as this is apparently targeted at that crowd.

SimonMW
Offline
Joined: Nov 16 2004
Quote:
as by Panasonic's resilience to come clean on this cam. For any professional broadcast/film use we need to know everything possible about our tools, warts and all.

I agree, and made the same point on DVinfo (although I haven't been back there to see if my comment has been removed!) Put simply Panasonic are making claims about this camera that simply aren't true. Jan Crittendon has a lot to answer for. She loves to dismiss me out of hand because I made a habit of calling out the marketing BS. But she is now on record as saying that Barry Green and herself shot charts that obtained completely different results. Amazingly these charts haven't surfaced for public examination yet. Funny that.

Quote:
Nobody comes out of cinema talking about the picture resolution.

That's true, but by the same token there are no horrendous aliasing issues with those projections either.

Gyr
Offline
Joined: Jan 17 2005
SimonMW wrote:
I agree, and made the same point on DVinfo (although I haven't been back there to see if my comment has been removed!)

It hasn't been removed, but the moderator/website owner has locked the thread. He says that was because it was going round in circles. I suspect that's only part of the reason. I believe he's embarrassed that the comments made by you and David Heath have been fully vindicated by the findings in Alan's report.

Panasonic haven't come out of this with any credit, but I doubt I'm alone in thinking that the credibility of DVinfo has taken a hit as well.

The earlier thread (Does the AF-100 line skip?) was locked with the moderators comment that "What difference is made by all the measurbating? Zero."

The truth is rather different. At the end of the day the testing ("measurbating") has shown that "the high level of aliasing in the original image prevents performance at 720p from being acceptable" and that "This camera does not perform particularly well at HD".

MAGLINK
Offline
Joined: Mar 8 2007

They did the same with the HPX370 debate last year when it all got too close to the truth!

Then bingo a firmware update and all talk of the new sensor disappeared, now there is a $2k rebate on the 370 so will we have a 470 this year and maybe an AF200 next year with a new super improved sensor (filter)?

SimonMW
Offline
Joined: Nov 16 2004

Chris runs that forum with an iron fist because he doesn't like offending his sponsors. While I understand that he doesn't want childish argument on there, he goes too far IMO, shutting down threads that are merely "vigorous debate". The forum lost credibility a long time ago because of this.

A classic moment was when I replied to a comment that someone made about DSLRs meaning that nobody needs to bother with lighting much any more. I pointed out that this was a load of rubbish, and as a result I was told by one of the moderators that I should be careful and shut up because "this isn't DV Opinion.net"!

There's the attitude to open debate they have right there!

Anyway, gone OT now, back to the 101.

MAGLINK
Offline
Joined: Mar 8 2007

This is the only decent un-biased forum I am on and I find DVI and DVX very NTSC related and full of people who are not prepared to respect other people's opinions, there are some sensible people there too but like me they tend to just walk away when the debates get silly.

As for the AF101 I personally am not keen on it having had a good look at BVE, I would get one if it was specifically requested for a job but I would rather use the RED and give the client proper HD, the problem is though that the DSLR shallow DOF thing has de-valued a lot of shooting here in the north east and there are all sorts of muppets out there who just go out and buy the camera and do it themselves, then they call me up to sort out all the problems in the edit, most of which seem to be footage out of focus!

MAGLINK
Offline
Joined: Mar 8 2007
SimonMW wrote:
Chris runs that forum with an iron fist because he doesn't like offending his sponsors. While I understand that he doesn't want childish argument on there, he goes too far IMO, shutting down threads that are merely "vigorous debate". The forum lost credibility a long time ago because of this.

A classic moment was when I replied to a comment that someone made about DSLRs meaning that nobody needs to bother with lighting much any more. I pointed out that this was a load of rubbish, and as a result I was told by one of the moderators that I should be careful and shut up because "this isn't DV Opinion.net"!

There's the attitude to open debate they have right there!

Anyway, gone OT now, back to the 101.

Just noticed my thread regarding Alan's report has now been closed, so much for free speech and debate in NTSC land? :rolleyes:

Gyr
Offline
Joined: Jan 17 2005

Bear in mind that Chris Hurd at Dvinfo locked it on a comment which he said was the “perfect concluding statement”.

That “perfect concluding statement” ended by saying that “The AF100 can be a great business decision and make very pretty pictures. And Alan Roberts has no bearing on that”.

I (and many others) regard Alan’s reports as impartial and objective factual reports and with regard to the AF-100 he noted that "the high level of aliasing in the original image prevents performance at 720p from being acceptable" and that "This camera does not perform particularly well at HD".

So take that “perfect concluding statement”, remove Alan’s name and replace it with a couple of key points from his report and hopefully people can see how questionable that “perfect concluding statement” was:-

“The AF100 can be a great business decision and make very pretty pictures and the fact that the high level of aliasing in the original image prevents performance at 720p from being acceptable and that it does not perform particularly well at HD has no bearing on that.”

MAGLINK
Offline
Joined: Mar 8 2007

I just await panasonic to bring out the AF200 next year with an improved sensor??? (filter) DVi and DVX will then be full of threads about why it sucks followed by a firmware update that gives them an AF100 again! ;0)

MAGLINK
Offline
Joined: Mar 8 2007

OOp's Simon I think we have rattled a few too many cages as I have been told that DVI has closed the AF100 section and it now links to the DVX page!

P.S Oh dear Jan's just closed her own DVX thread on the good reports for the camera as well.

SimonMW
Offline
Joined: Nov 16 2004

I've seen Chris overreact to things before, but that is pretty extreme! It is clear from DVX User that there are a lot of disappointed people out there.

MAGLINK
Offline
Joined: Mar 8 2007
SimonMW wrote:
I've seen Chris overreact to things before, but that is pretty extreme!

I don't think I have ever seen anything quite like all of this regarding a camera and all the negative reports on the DSLR's didn't seem to affect their sales, maybe after NAB it will all be a different story? :confused:

Piotr
Offline
Joined: Jun 9 2006

DVXuser just closed the "AF100 BBC tests" thread too. "You asked for it" the mod says. This is madness. Because of a ...camera!?
I haven't long seen anything as mindless as that from the home of the free (at least in the area of exchange of opinion).

MAGLINK
Offline
Joined: Mar 8 2007

Well Barry Green just went way down in my respect list after that final post on DVX, do we all have to accept everything that the god panasonic gives or tells us, Jan seems to be a bit out of her depth too if she can't take the flak from her customer base or more importantly potential future customers!

I used to work for AMS Neve as it's main operational customer support manager and I would never dismiss customer's complaints about current or new products in this manner even if they were really bad, I would just take them in and see if R&D or us as a company could fulfil the customers requirements or queries.

SimonMW
Offline
Joined: Nov 16 2004

He went down in my estimation a long time ago. Another guy with a forum who is protecting his sponsors. They don't seem to get it that a multi national electronics company and the reps that are paid to represent them like Jan, with their large guaranteed salaries, are perfectly capable of defending themselves.

The trouble is that guys like Barry are enthralled that a company like Panasonic graces his forum directly. Apparently according to Barry, if you are a person that complains about his closing of the thread you should go away. So I suppose then that customers who bought the 100 thinking they were getting an HD camera and who are now disappointed cannot take their gripes to a public forum where Panasonic might be forced to give some answers.

But hey, at least he gets to protect his ability to get hold of the next Panasonic prototype to play with...

Quote:
I used to work for AMS Neve as it's main operational customer support manager and I would never dismiss customer's complaints about current or new products in this manner even if they were really bad, I would just take them in and see if R&D or us as a company could fulfil the customers requirements or queries.

Panasonic has always had a defensive streak. They do not like any sort of criticism. I'm sure that if they owned attack dogs with lasers on their heads they would send them after you if they took a disliking. If only you knew about the debacle when a reviewer for the now defunct Showreel Magazine once made a tiny, tiny criticism of the HPX-800 when he reviewed it. It resulted in a very public row with the editors, and the reviewer was banned by Panasonic from reviewing their gear ever again. All he did was air a slight doubt as to the long term viability of P2. He didn't make anything of it. Just a slight aside from the main review. Yet they went ballistic.

I had to get hold of my own HPX500 to review because they wouldn't let me near one due to my posting on forums about my doubts about P2 at the time.

Then there was the issue of fake blogs paid for by Panasonic but made out to be independent that people found out about. Panasonic, quite frankly, have a crap record when it comes to PR and dealing with issues.

MAGLINK
Offline
Joined: Mar 8 2007

Oh well at least everything is happy in La La land as NAB is coming soon, no doubt then we will have them all in glorious video saying how wonderful it all is and they will sell shed loads of them, then it will all start again like it did with the HPX300 and hey next year the AF200 will have its much improved new "hyper gamic endosensor" "tm" with added depth and punch so that you can shoot in even darker rooms with +18db's of gain.

In the meantime back at panasonic HQ sponsor of the London 2012 Olympic games many heads are scratched as to why the HD camera's with tapes inside them shot some of the best ever BBC footage and now everyone seems to be using XF305's and Sony 350's and 500's;)

Or even the new Sony Hasselblad :D

infocus
Offline
Joined: Jul 18 2003
JGNattrass wrote:
I don't think I have ever seen anything quite like all of this regarding a camera and all the negative reports on the DSLR's didn't seem to affect their sales, maybe after NAB it will all be a different story? :confused:

Ah, but DSLRs are first and foremost still cameras, and the video mode was marketed for them as a bit of an extra. It was then users who took the initiative, and started to use them for purposes they were never primarily promoted for. There is also the question of price - look how much a DSLR costs, and compare it to any true video camera. I recently bought a Canon 550D - but I was fully aware of what I was buying and it suits it's purpose well. But I wouldn't try to pass it off to a client as rivalling a "proper" HD video camera.

This is different. The 101 was promoted as a professional digital cinema camera ("Optimised for high-definition video recording, the large image sensor on the AF101 enables users to achieve cinema-like results") - in practice that is a big overstatement. It would be wrong to describe it as a bad or faulty camera - as far as I know, none have so far exploded in use ;) - but it is not up to the standard that had been claimed for it.

rone01
Offline
Joined: Feb 9 2009

With regard to closing threads, I don't think it's about taking flak or criticism it's about the way things can't actually be concluded on forums and the internet in general - where often a reasonable debate just goes into choas quite quickly - hence the mods have no choice. DVX is a very popular community and is naturally biased - it's called DVXuser for pete's sake! There are lots of successful and interesting people on there, the production world is a better place for it - (even had a couple of clients from it - one a released feature using HVXs). I've moved through the times with DVX,HVX,HPX,LETUS and now RED - just because I started reading there!

End of the day, it's only a camera and if you like the thing and make money from it - like everything in life that's all that matters. BBC HD standards may mean something to some people but not to all and that's where there is a huge disproportion of malaise, rivalary and snobbery is born. The tests are one useful element in a considered purchasing decision and to be honest have never been a part of my decision as I don't do HD for the Beeb! (Apart from the odd bit) My decisions have been made on the demand of clients and my production workflow.

Everything is fluid in life, nothing is certain do what you want to do with a camera.

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

Piotr
Offline
Joined: Jun 9 2006

Well, I do only HD for my clients, including nationwide broadcasters , Disney, and such, and they go by the EBU/BBC recommendations. End of story. Each to his own business.

rone01
Offline
Joined: Feb 9 2009

That's fine and good with it too. It's all relative. We have a low budget contingent (P2 based) and higher budget operation (RED) so we cover the bases we need.

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

Gyr
Offline
Joined: Jan 17 2005

These internet storms arise about all sorts of products from time to time and I’m coming to the conclusion that a manufacturer/supplier/retailer can normally ride them out (and even benefit from them) IF the basic facts are on their side.

That being the case I think Panasonic may be in for a bumpy ride, though the fact that Dvinfo and DVXUser won’t take any negative comments may at least partially shield them.

In contrast I saw a recent case where a customers online rant and call for others to boycott that firm led to the posting of a succession of very positive testimonials for the firm in question.

johnmercer
Offline
Joined: Jan 12 2008

Wow! I was one of the people posting on that thread along with Simon and Gary. This kind of censorship will only bite them in the ar__ I feel. I was trying to be even handed but I did make a tame reference to Jan and Barry Green, so maybe it was me :confused: I don't think anyone can reasonably argue with Alan's professional science and methodolgy and lack of axe to grind, and so they close it down! Very childish if you ask me. I won't go to any of those forums now, stuff 'em.

What is the point now of talking intelligent shop on the net when you have rabid fanboyism of Reduser, DVXuser and now DVinfo.net? 99.9 % of these people will never get beyond a test on Vimeo and I mean absolutely no disrespect to the thousands of prosummers, corporates or independents and students out there but why do the maunfacturer's marketing for them, for free? Dog and tail wagging, and all that. But also if technology is going to be driven by fanboy hysteria and dreams then it makes it harder ultimately to get the right no nonsense tools.

There has to be a measuring by the highest professional broadcast standards otherwise we and the public will be duped. There is a place for all these production tools and some people will always make great stuff but these manufacturers are making and selling gear to a dream and sometimes glossing over the reality that it takes more than a toy to make the kindergarden. Just my 2 pennies and happily see that someone else views it differently.

infocus
Offline
Joined: Jul 18 2003
rone01 wrote:
End of the day, it's only a camera and if you like the thing and make money from it - .....

Yes.... but I think there is a point you're missing.

If the controversy was solely about how good the camera was, then I'd agree with you. But the trouble goes much deeper. It appears that official claims were initially made as to performance, those were disputed by users, then those users were categorically told they were wrong. Now it turns out that certain key claims made by the manufacturers were wrong - the users were right all along. The complaints are not just about the performance per se - far more about a perception of being misled.

Just think of some classic political scandals. Whether a minister does or does not have a mistress may not matter a jot regarding how they run the country (though it may matter a great deal to their family). BUT, if they were publicly proclaiming family values, they lay themselves open to charges of hypocrisy. If they were to initially deny the allegations, they lay themselves open to charges of dishonesty.

Both of those are far more damaging in the public eye than just infidelity. In lots of matters, the issue itself may be far less important than how an individual or corporation deals with it.

MAGLINK
Offline
Joined: Mar 8 2007

I was in the process of testing the camera myself but unfortunately a very good looking post-person rang the doorbell, I was about to show the results of my zone chart when I found myself in bed with said post person, I apologise to the reader's of DV Doc for my lack of judgement in this matter and to my wife who understands that testing camera's to this level can cause this sort of thing to happen!;)
My conclusion in all of this sorry matter is that it is not only the Af101 that sucks!!!

johnmercer
Offline
Joined: Jan 12 2008
Gary Nattrass wrote:
I was in the process of testing the camera myself but unfortunately a very good looking post-person rang the doorbell, I was about to show the results of my zone chart when I found myself in bed with said post person, I apologise to the reader's of DV Doc for my lack of judgement in this matter and to my wife who understands that testing camera's to this level can cause this sort of thing to happen!;)
My conclusion in all of this sorry matter is that it is not only the Af101 that sucks!!!

We have the fullest confidence in Mr. Nattrass and will stand by him at this unfortunate time (until we stab him in the back, in a two faced manner, and sack him from testing cameras) and we look forward to him returning to testing of more appropiate objects very soon. We also note that Mr. Natrass comments about the AF101 were made under some personal duress and we enjoy the utmost of cordial diplomatic relations with the AF101.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

They don't need to improve the sensor in the AF101, that's just fine. They need to improve the down-scaling algorithm that generates 1920x1080 from it, and to include a decent optical low-pass filter, then it'll truly be the dog's danglers. And I assume that, if you've read my report, you'll have noticed that those are my only comments against it.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Gyr
Offline
Joined: Jan 17 2005
Alan Roberts wrote:
They don't need to improve the sensor in the AF101, that's just fine.

As I understand it the forthcoming Sony NXCAM will probably use the sensor from the F3, which at about 4 megapixel, is (in theory at least) a better size for HD video than the sensor in the AF101. However I'm happy to be corrected if that is wrong.

SimonMW
Offline
Joined: Nov 16 2004
Quote:
And I assume that, if you've read my report, you'll have noticed that those are my only comments against it.

That, and the performance. Let's not make any bones about it, as Infocus mentions, whether the performance is good or bad is irrelevant in a wider context. What makes this a real issue is that Panasonic are on record as saying that the camera not only resolves a good 800 lines, but also that all the aliasing issues of DSLRs have been eliminated. While they have removed the rainbow moire that plagues DSLR footage, they haven't, from what I can see, done any of the other things that they claim of it.

LPowell
Offline
Joined: Mar 17 2011

My first post here at DV Doctor, though some of you may have noted my speculations on AF100 judder that Piotr reprinted from a thread at dvxuser. By way of introduction, I'm an audio DSP engineer currently employed at Dolby Labs, and an amateur videographer. At dvxuser, I'm also fairly well known for a series of GH1 patches I developed, most notably my Reliable In-Camera Playback Patch.

I found Alan's detailed tests of the AF100 very interesting, and posted an interpretation of his results in this dvxuser thread, before it was closed:

http://www.dvxuser.com/V6/showthread.php?241960-AF101-BBC-EBU-test-results-posted/page13

I had a few more questions for Alan, regarding the 1080psf luma zone plate on page 9 of his report.

1. While the horizontal and vertical high-resolution aliasing circles are quite prominent, there appears to be almost no diagonal aliasing. Is this due to the cross-shaped geometry of the sensor demosaic filters? Or perhaps diagonal aliases occur a bit beyond the outer edge of the 1080 circle?

2. In the absence of aliasing, clean diagonal resolution appears to extend out to about 80% of the diameter of the zone plate, which would be around 860 lines. Can we conclude that this is the base resolution of the sensor when free of aliasing artifacts? And that it would have 860 lines of base resolution horizontally and vertically as well?

3. With the red and green zone plates, you observed that the green signal appears to have been derived from the sensor at the same resolution as red and blue signals. Would this explain the absence in the AF100 of the type of color-tinged aliases that plague the Canon 5D Mark II?

4. Would the addition of an OLPF steep enough to block the horizontal and vertical aliases out to 800+ lines resolve the camera's image quality issues in your view? Or does the aliasing beyond 800 lines need to be cleaned up as well?

rone01
Offline
Joined: Feb 9 2009
infocus wrote:
Yes.... but I think there is a point you're missing.

If the controversy was solely about how good the camera was, then I'd agree with you. But the trouble goes much deeper. It appears that official claims were initially made as to performance, those were disputed by users, then those users were categorically told they were wrong. Now it turns out that certain key claims made by the manufacturers were wrong - the users were right all along. The complaints are not just about the performance per se - far more about a perception of being misled.

Just think of some classic political scandals. Whether a minister does or does not have a mistress may not matter a jot regarding how they run the country (though it may matter a great deal to their family). BUT, if they were publicly proclaiming family values, they lay themselves open to charges of hypocrisy. If they were to initially deny the allegations, they lay themselves open to charges of dishonesty.

Both of those are far more damaging in the public eye than just infidelity. In lots of matters, the issue itself may be far less important than how an individual or corporation deals with it.

You can't really compare a political scandal to a bit of marketing spin can you? The language here is wee bit strong.

Somehow I can't see John Hurt reprising his Profumo role this time taking the path of Panny's 800 line boast to court?

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

infocus
Offline
Joined: Jul 18 2003
rone01 wrote:
You can't really compare a political scandal to a bit of marketing spin can you? The language here is wee bit strong.

The intent of my analogy was not intended as a reference to the seriousness of the matter, as much as the general principle: ".........the issue itself may be far less important than how an individual or corporation deals with it". I apologise if I didn't make that clear.

In this case, the actual performance of the camera is less relevant than the fact that false claims seem to have been made for it.

rone01
Offline
Joined: Feb 9 2009

I would imagine in terms of priority Panasonic-Japan-Osaka have got their hands full at the moment - that said Jan from the US said they will respond.

As an aside where is the 800 line quote? Anyone know?

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

infocus
Offline
Joined: Jul 18 2003
LPowell wrote:
My first post here at DV Doctor, .......

Welcome!

LPowell wrote:
In the absence of aliasing, clean diagonal resolution appears to extend out to about 80% of the diameter of the zone plate, which would be around 860 lines. Can we conclude that this is the base resolution of the sensor when free of aliasing artifacts? And that it would have 860 lines of base resolution horizontally and vertically as well?

Somebody else bought this up on another thread in this forum, and it may be worth thinking about in more detail.

It is quite normal to expect greater diagonal detail than horizontal or vertical, and the explanation I was offered is that it's because it is a resulatant partly of the horizontal photosite count, partly of the vertical. Mathematically, it's a vector resultant.

Hence, in this case, the diagonal resolution would be expected to be 1.414 (sq rt 2) times that of the basic horizontal and vertical resolutions.

Since basic resolution here seems to be 661, the expectation for the diagonal would be verging towards about 930 lines. Allowing for expected losses, that correlates very closely with the 860 you refer to.

If you doubt any of that, look at the zone plate for a camera like the V1, where the photosites are arranged on the diagonal. Lo and behold, the diagonal resolution is there LESS than vert/hor.

LPowell wrote:
3. With the red and green zone plates, you observed that the green signal appears to have been derived from the sensor at the same resolution as red and blue signals. Would this explain the absence in the AF100 of the type of color-tinged aliases that plague the Canon 5D Mark II?

In the other thread ( http://forums.dvdoctor.net/showthread.php?p=382402#post382402 ) I copied a note from a friend questioning some of Alans conclusions. (Though not the fundamental measurements.) It may be worth repeating it here, as it answers your question well. I'd be interested to hear your comments about it.

The 5D issue seems due to that process skipping whole lines in the read process, and binning charge even within the read lines. The AF101 method outlined below is still a pixel skipping process, but a neater way of doing it.

------------------------------------------------------------------------

There is a method of reading the sensor which could involve pixel skipping, but would not give any colour aliasing. To my mind, it's a more likely explanation for measured results than reading at high resolution and downscaling.

It will also give equal R,G,B resolution, each at one quarter the native sensor resolution and would mean reading in the following manner:

<------------------4700 photosites----------------- >

o o R G o o R G o o R G o o….. ^
o o G B o o G B o o G B o o….. 2644 photosites
o o o o o o o o o o o o o o……… total vertically
o o o o o o o o o o o o o o…..
o o R G o o R G o o R G o o…..
o o G B o o G B o o G B o o…..
o o o o o o o o o o o o o o………
o o o o o o o o o o o o o o…..

Where R, G, B are read red, green, blue photosites, and o are skipped ones.

The basic unit becomes :
o o R G
o o G B
o o o o
o o o o

And it should be obvious that there will be a quarter the no of those (horizontally and vertically) compared to sensor photosites. Hence, basic camera resolution defined by the number of basic units, 4700/4 and 2644/4, or 1175x661 – exactly as measured.

Such an explanation would :

Give no colour aliasing
Give equal resolution for each colour (and each at 1175x661)
Explain the sensitivity/noise
Be consistent with low rolling shutter
Be simple to implement (it’s probably what the GH2 uses so no extra R&D)

All consistent with what is measured.

What makes me favour such an explanation? Have you ever heard of “Occams Razor”? It’s a guiding principle in science which roughly states “…..we should tend towards simpler theories until we can trade some simplicity for increased explanatory power”

In this case, the razor points towards an explanation similar to above, and away from any explanation that suggests anything far more complicated, and a poor downconversion.

It's by far the simplest explanation which fully explains measured results.

-------------------------------------------------------------------------------------------------------

Gyr
Offline
Joined: Jan 17 2005
rone01 wrote:
I would imagine in terms of priority Panasonic-Japan-Osaka have got their hands full at the moment - that said Jan from the US said they will respond.

As an aside where is the 800 line quote? Anyone know?

I don't know if it's in any of Panasonics literature, but on one of the now deleted threads on DVinfo Jan said "I know that my engineer, Barry Green and I all saw 800 lines".

She also said there was no aliasing on her charts which isn't exactly compatible with the test results Alan obtained ("the high level of aliasing in the original image prevents performance at 720p from being acceptable").

Lachrymology
Offline
Joined: Mar 16 2011

It's in the manual vol 1 pg 48, "horizontal resolution - 800 TV lines (standard, in the centre of the screen)"

So trying to justify it by saying it's 800 on the diagonal doesn't cut it (which it does appear to resolve on Barry Green's Wringer, boy that sounds wrong).

rone01
Offline
Joined: Feb 9 2009

Hmm, interesting.

A deleted internet thread with a Panasonic employee claiming they have observed 800 lines is hardly worth hanging on to - I would be more concerned it it was hard-copied onto a brochure with Panasonic's brand all over it.

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

Piotr
Offline
Joined: Jun 9 2006

eg. AF100 manual (v1) specs on page 49:

"Horizontal resolution
800 TV lines (standard)
(in the center of the screen)"

johnmercer
Offline
Joined: Jan 12 2008
Alan Roberts wrote:
They don't need to improve the sensor in the AF101, that's just fine. They need to improve the down-scaling algorithm that generates 1920x1080 from it, and to include a decent optical low-pass filter, then it'll truly be the dog's danglers. And I assume that, if you've read my report, you'll have noticed that those are my only comments against it.

Do they really want to I wonder? What is the commercial incentive (apart from perhaps misjudging this particular release I mean)? It is my view that these cameras are made for a particular market and that is the prosummer shallow DOF independent sector which has been very lucrative for the DSLRs. To make them so good as to rival other Panasonic products or F3s would have an unpredicatable commercial impact that may be beneficial in the short term but counter-productive in the long term. This I believe could be the real reasons behind the repeated delays to market of the fixed lens 3k Red Scarlet. It has to respond to much cheaper DSLRs and now AF101s and else be clearly demarced from the higher end Epic. Its brief and technical aspects/pricepoint has clearly been thought of very carefully whilst watching the market.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

1: The identical H and V resolutions are due to the sensor having the same resolution H and V, equal in each of R G and B. I suspect this means that the Bayer pattern is decoded separately at the sensor resolution for each of R G and B, and then each downscaled. I thibk I said that in the report.

2: No, you always get more diagonal resolution in a camera with orthogonal rectilinear resolution, i.e where the H and V resolutions are not interdependent. In this respect, it's just like any normal 3-sensor camera. which is a welcome change.

3: yes, and I think I said so in the report without specifically mentioning any other camera (I don't do comparisons).

4: they need to do two things

4a: improve the downscaling from sensor resolution to video resolution by using better (i.e. much bigger and therefore expensive to operate) filters. This will drop the level of aliases, and can, in the extreme, eliminate them entirely, but the filtering operation gets quite hard to do. I have filters which do exactly that, software, my design, so I know it can be done, but the number of terms in the FIR s=design gets spectacularly big.

4b: Optical low-pass filters don't have a sharp cut-off, that's beyond the realms of physics, the roll-off is quite smooth, steeper than the Sinc function of the camera itself, and steeper than the graceful droop of the lens, but still not steep enough to impose a sharp cut, that can't be done. the fix is to reduce the cause of he aliasing (the high frequencies reaching the sensor) with an optical filter, and to reduce the appearance of what's left using better electronic filters in the scaling.

There's no single fix, that's why it's hard, and expensive, to get right.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

MAGLINK
Offline
Joined: Mar 8 2007

As we know all camera's are built to a cost these days so to expect the AF101 to be a RED is probably asking too much, that said though if it is being sold as a full 1080 HD camera then I would expect the performance to be somewhere near that.

As for filters they did that to the HPX301 last year and called it the 371, adding the filters caused all sorts of other problems that were not there on the original camera so a firmware release was done which basically gave you a 301 again. There were claims of a new sensor but that did not seem to be true in the case of the 371.

So at the end of the day I suppose it is a trade off between resolution and problems that need expensive solutions so as an user we just have to make the choice that suits our needs.

infocus
Offline
Joined: Jul 18 2003
Alan Roberts wrote:
4a: improve the downscaling from sensor resolution to video resolution by using better (i.e. much bigger and therefore expensive to operate) filters. This will drop the level of aliases, and can, in the extreme, eliminate them entirely, but the filtering operation gets quite hard to do. I have filters which do exactly that, software, my design, so I know it can be done, but the number of terms in the FIR s=design gets spectacularly big.

I've now been offered two possible explanations for how the results are obtained. Alans, and the one I previously copied.

On the face of it, the one I copied (at post 44) seems far more plausible to me - it's a far simpler explanation which still fits with all the known facts. In fact, it's actually MORE likely in terms of explaning some facts such as low rolling shutter. Fewer pixels to read, so faster readout.

Of course, I may be missing something fundamental, and therefore I would be very grateful, Alan, if you could at least give it due attention, and if you truly believe it cannot be correct, explain why you think that to be the case.

As far as your words above go, then whilst poor filtering may explain away a high level of aliasing in the end result, then why the low resolution? Surely even the poorest downconvertor should be able to get a better result than that, aliasing issues aside?

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I don't see why I should have to repeat explanations. I've already given the measurements and maths in the document, and they're consistent. The science is fairly obvious, and I see no reason at all for wasting time on continually restating it.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

infocus
Offline
Joined: Jul 18 2003
Alan Roberts wrote:
I don't see why I should have to repeat explanations.

Alan, the request was to take a look at the alternative explanation, and if you can find any fault with it, to explain why it's wrong.

Quote:
I've already given the measurements and maths in the document, and they're consistent. The science is fairly obvious, and I see no reason at all for wasting time on continually restating it.

Quite a few people find one strong inconsistency in the document, which has already been mentioned.

You say in the report "The horizontal and vertical aliases have a centre frequency of 1.224 times the system frequencies. This implies that there is an underlying resolution 1.224 times that of 1920x1080, or 2350x1322..."

Surely centres of the alias circles point to frequencies TWICE the underlying resolution, not the underlying resolution itself?

Hence an underlying resolution of 1175x661 - not 2350x1322. I've now checked this with a number of people, who are all in agreement, and would be grateful if you could address this one specific point directly.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

No more, I've had enough, I've got FAR better things to do than to argue irrelevancies. I've got REAL WORK to do, earning money, I haven't the time to waste in these stupid arguments. Get a life, get a job, go and do something useful.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

SimonMW
Offline
Joined: Nov 16 2004

Sorry, Alan, but your reaction is totally uncalled for. All that was asked of you was to entertain the idea of an alternative explanation, and explain why it wasn't doing it that way instead. If you have direct information from Panasonic stating that it does things in the way you explain in the BBC report, then fine. If your reasoning is pure maths, then it stands to reason that Infocus's theory is also logical also.

A man of your knowledge should be able to answer such a query in a straightforward manner instead of getting the hump and throwing your toys out of the pram.

LPowell
Offline
Joined: Mar 17 2011
infocus wrote:
There is a method of reading the sensor which could involve pixel skipping, but would not give any colour aliasing. To my mind, it's a more likely explanation for measured results than reading at high resolution and downscaling.

It will also give equal R,G,B resolution, each at one quarter the native sensor resolution and would mean reading in the following manner:

<------------------4700 photosites----------------- >

o o R G o o R G o o R G o o….. ^
o o G B o o G B o o G B o o….. 1322 photosites
o o o o o o o o o o o o o o……… total vertically
o o o o o o o o o o o o o o…..
o o R G o o R G o o R G o o…..
o o G B o o G B o o G B o o…..
o o o o o o o o o o o o o o………
o o o o o o o o o o o o o o…..

Where R, G, B are read red, green, blue photosites, and o are skipped ones.

The basic unit becomes :
o o R G
o o G B
o o o o
o o o o

And it should be obvious that there will be a quarter the no of those (horizontally and vertically) compared to sensor photosites. Hence, basic camera resolution defined by the number of basic units, 4700/4 and 2644/4, or 1175x661 – exactly as measured...

It's by far the simplest explanation which fully explains measured results.

While it is quite simple and may correlate well with some measured results, this sampling kernel wouldn't come anywhere close to matching the perceived image quality of the AF100. Sparse matrix sampling is highly susceptible to randomly-aligned chroma and contrast variations in an image, and would produce much cruder ragged edges than we see in AF100 frame samples.

Other issues are that this theory implies the sensor has only 6.2 Megapixels and that it does in fact use line-skipping. It's asking us to believe that Panasonic is not only technically incompetent, but lying about the product as well.

Duncan Craig
Offline
Joined: Nov 19 2008

I'm with Alan here, he's not obliged to answer anything.
The fact that the other forums have given up on the subject doesn't mean everyone should come here to air their grievances.

SimonMW
Offline
Joined: Nov 16 2004
Quote:
I'm with Alan here, he's not obliged to answer anything.
The fact that the other forums have given up on the subject doesn't mean everyone should come here to air their grievances.

I didn't say he was obliged to answer. I just said that a person of his knowledge should be able to explain an answer to the questions made without acting like a spoilt teenager. Alan dismissed a technical question with theory behind it as an "irrelevancy". Sorry, but that just doesn't cut it I'm afraid. Infocus wasn't being argumentative, but asked a question. Yet Alan has said it is is being argumentative. Utter, utter nonsense. They were questions, not arguments. If Alan deems anything that might contradict him as an argument/threat then I think he has a few issues going on.

Quote:
It's asking us to believe that Panasonic is not only technically incompetent, but lying about the product as well.

The second part you may be on to something (ie look at the results of resolution and aliasing, then compare to the Panasonic brochures). At the very least they were being economical with the truth. It doesn't resolve 800 lines as they claim, and it doesn't eliminate aliasing. So guilty on both counts I would say.

infocus
Offline
Joined: Jul 18 2003
LPowell wrote:
Other issues are that this theory implies the sensor has only 6.2 Megapixels and that it does in fact use line-skipping.

No, the theory suggests 1175 blocks of 4x4 photosites horizontally, and 661 blocks vertically. Hence 1175x661 blocks (776,675) of 16 photosites each - so 776,675x16 photosites in total, or 12,426,800. Effectively 12.4 megapixel. [EDIT Whoops, and well spotted! There is an error on the diagram - it should, of course, read 2644 photosites vertically, not 1322. My apologies. I'll amend the original.]

Line-skipping? Well, sort of. But not in the way some DSLRs do it which is very coarse, and will inevitably give an assymetrical result. This theory suggests a symmetrical result. (As observed.) Far better to call it pixel skipping.

LPowell wrote:
It's asking us to believe that Panasonic is not only technically incompetent, but lying about the product as well.

Again no. This would NOT be at all technically "incompetent". It doesn't give the result for resolution/aliasing that a lot of people were expecting - but it does give a better result than many DSLRs for video. And, crucially, it is likely to be quite cheap and simple to implement, and make for low power consumption and fast sensor reading. Possibly even making use of GH2 processing, as well as the basic sensor.

On the other hand, reading the sensor at half resolution (ie. as 2x2 blocks), to get a 2350x1322 resolution result, then downscaling so badly to only end up with such a low result...... That really would be technically incompetent.

And "lying" implies a deliberate deception. But what about a scenario where the senior technical people give a presentation along the lines of "our camera does not use simple line skipping such as a DSLR, it has a much better way of reading the sensor to avoid the colour aliasing". By the time the time the message has been passed down the chain, it's got turned in to "it doesn't line skip or pixel skip". No deliberate deception - just unfortunate misunderstanding. "We're going to advance" getting turned into "we're going to a dance".

LPowell wrote:
Sparse matrix sampling is highly susceptible to randomly-aligned chroma and contrast variations in an image, and would produce much cruder ragged edges than we see in AF100 frame samples.

A good question. But what about if the AF101 DID have an OLPF, but with quite a high cut-off frequency such as not to lower the desired resolution even more. Say one having effect more around the horizontal 2500 line mark? So enough to "smudge" the detail over the 2x2 photosite blocks, but not enough to affect "real" resolution, that up to 1175x661?

Which would also mean Panasonic were telling the truth about an OLPF. The cut-off may not get rid of all the aliasing we would like, but maybe it would nullify the effects you refer to?

Look, I don't know the definitive answer, I freely admit that, and I doubt anybody outside Panasonic does for sure, but I still feel something along these lines is the most likely explanation. There are all sorts of variants, such as reading all 16 photosites and binning the charges, but I do feel any explanation involves dealing with 4x4 blocks as the basic unit.

But thanks for bringing up those points. They are worth thinking over, and thanks for pointing out the typo.

LPowell
Offline
Joined: Mar 17 2011
Alan Roberts wrote:
1: The identical H and V resolutions are due to the sensor having the same resolution H and V, equal in each of R G and B. I suspect this means that the Bayer pattern is decoded separately at the sensor resolution for each of R G and B, and then each downscaled...

2: No, you always get more diagonal resolution in a camera with orthogonal rectilinear resolution, i.e where the H and V resolutions are not interdependent. In this respect, it's just like any normal 3-sensor camera. which is a welcome change.

Thanks for the detailed explanation, it not only accounts for the similarity of the R and G zone plates, but reveals why the AF100 cannot also function as a high resolution stills camera like the GH2.

Actually, I think it's a pretty innovative design - emulating an HD-resolution 3-CCD sensor with a single high-resolution MOS sensor. Even with the AF100's aliasing problems, its downscaled patterns of same-color photosites produce subjectively better-looking images than the color-fringed results of using a demosaic filter on a subset of the photosites (e.g. Canon 5D Mark II).

It occurs to me that perhaps the AF100 is a rather clever compromise between high resolution and low cost. It uses an inexpensive OLPF that passes just enough high frequency details to produce aliased artifacts that artificially augment the perceived resolution of the sensor. It then compresses these images with an encoder whose low bitrate tends to smooth out the worst of the artifacts. Improving the quality of any one of these components - the OLPF, sensor resolution, or encoder bitrate - would risk exposing the deficiencies of the other two. They are instead fine-tuned to pleasantly conceal each others' flaws.

Hence, a video camera that looks much better than it tests.

Duncan Craig
Offline
Joined: Nov 19 2008

By continually questioning someone's competence and questioning their work you are almost goading them into answering you. At the end of the day Alan has published his work and if Panasonic had anything to say about it they would have done so.

I'm grateful someone like Alan takes time to help people on here, it must be a labour of love for him. If he feels annoyed by folks on here, it might not be because he doubts his own work, simply that he's pissed off with the constant agro.

If you don't like his approach try publishing your own documents.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I apologise for my outburst last week. It was unduly vitriolic.

Here's the background.

For 11 years, I have been here, posting answers and solutions to problems, providing the science and psychophysics behind the science. Thousands of times. Many times, I have given the same explanation, largely because people seem unaware of the rather splendid search facilities here. Never mind.

Some of the questions have been interesting, some banal, needing only common sense for an answer. I've been happy answering them though, because it's always a pleasure seeing the dawning of understanding. Other forums seem not to work this way, being largely platforms for vested interests to shout at each other. That's why I stay here.

But, in the last couple of years or so, things have been changing. Now, the questions are more along the lines of 'you've got it wrong because...', and some of the posters and postings are clearly biased for/against individual cameras and/or manufacturers.

I have always been totally neutral on such things, I measure and report. My neutrality has often been questioned; I've been accused of being in the pay of both Sony and Panasonic, so I suppose my neutrality is still observable. It will always remain so.

The recent furore over the AF101 has been the last straw. I got so angry with the continued blether about it that the fuse finally blew. So, here's my statement for the future:

I WILL NO LONGER READ, OR REPLY TO, ANY COMMENTS ON ANY CAMERA, FOR ANY REASON.

If you want to talk rubbish about cameras, get on with it, I won't comment either way.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

johnmercer
Offline
Joined: Jan 12 2008

Alan I for one totally understand your reasonings but it's an absolute shame that it has come to this. I thank you for what I've always recognised as all the totally thorough scientific and unbiased testing, comment and feedback you've made and although I don't come here often it is usually to see what you have to say.

I think it's almost totally useless now to forage forums for reliable relevant and professional advise from real professional users. I will however always look out for your reports and hope they long continue.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I'll still be here for the science and psychophysics, but not for comment on cameras or manufacturers.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Philip Lipetz
Offline
Joined: Mar 15 2011

Alan, your reports are the ONLY decent source of information on the web. Other discussion forums are run by people whose economic interest lies in promoting a particular advertiser's products. Unfortunately, this produces a "fanboy" mentality that is not based on observation but aspiration.

What is happening in the AF100/101 thread is a spill over from other forums. Unfortunately, many of the leading forums have closed their AF100 threads to any whisper of rational examination. The same goes for other forums and other camera discussions.

While this is only my second post, DVDoctor has long been my refuge from the noise that makes forums into advertising rallies. When I talk to industry pros and mention your name the respect that you have is unbelievable. It almost like I get status just by recognizing your value and contribution.

I fear that your decision to withdraw will totally throw all of the discussion to forums that let advertising determine discussion, and will not allow any opinion that endangers advertisers. That would be a horrible world, worse than no discussion at all.

I know that the vitriol of this thread hardly reflects the reserved conversations that you may have been accustomed to during your years at the BBC. It hardly reflects my experience when interviewed at the BBC. However, distasteful it might be you should realize that you are the most significant voice of reason in this industry, and that your withdrawal from public discussion will throw dispassionate discussion under the wheels of the runaway advertising bus that the Internet is becoming.

I beg you to reconsider. Your position is so great that your presence is more than a personal matter, it is a matter of public welfare. Please see how it will affect the entire community of people, a community that goes well beyond the members of this forum because your words are repeated by all who respect your lifetime of effort.

MAGLINK
Offline
Joined: Mar 8 2007

I personally have had technical advice and knowledge from Alan and use his reports to make my decisions regarding the best camera kit to buy, I am very grateful for his expertise and unbiased reports.

I read his HPX301 report again two years after I bought one and I can now fully understand all of it and how the same testing was applied to other camera's and the AF101.

We are so lucky to have such free discussion's on this forum and long may it continue, I questioned several aspects of the AF101 myself on the other forums but if I am not allowed to voice my opinions or ask valid questions then I will leave those to the people who trust everything their chosen camera manufacturer tells them.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I've just had enough of the spurious arguments and questioning, particularly when the 'new theory' is completely bonkers and already dealt with in the test report. I'll be happy to talk one-to-one, but I'm not going to get drawn into arguments here, life's already hard enough, my BP's at stake.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

nattt
Offline
Joined: Aug 14 2000

In my experience Alan, your reports are not just well written and technically well performed, but are indeed unbiassed. I've learned a lot from reading them!

Graeme

nattress.com - Filters for FCP & Color
red.com - 5k Digital Cinema

Erichus
Offline
Joined: Aug 26 2009

I also read the forum posts here with daily interests, especially those from Alan, and I think it would be a pity, if Allan withdrawed from his lively and much appreciated comments. This as far from my part from Switzerland

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I'm happy to deal with PM's, but not with public discussions. You wouldn't believe how many crackpot theories I come across in a working year, and, were I to spend the necessary time to address them all, I'd not get anything else done. As it is, I'm currently working about 4 days a week (very long days), measuring things, giving university courses, training people and so on. I simply haven't the time or inclination to peruse and demolish.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

RCJardin
Offline
Joined: Jun 24 2001

Alan we appreciate the time and experience you do give us. Your knowledge is priceless.

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

And there's the rub, it comes at a price. In the past 3 weeks, I've had approximately 2 days not working, and I get paid rather nicely for doing it. If there's a choice between doing that and spending the hours needed to debunk the loony theories, I'll take the money any time.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

dude
Offline
Joined: Nov 2 2010
Alan Roberts wrote:
For 11 years, I have been here, posting answers and solutions to problems, providing the science and psychophysics behind the science. Thousands of times. Many times, I have given the same explanation, largely because people seem unaware of the rather splendid search facilities here.

Just so you know, I for one, even if I’m new to forum, just keep looking for your past posts, reading it (and saving it) with your (and others) arguments or opinions. It's a lot of broadcast history (as with EBU 720p 'decision'), Planet Earth (very interesting!) stuff which help me to understand how ‘we’ came a long way and, obviously, a lot of tech - real gold for me. Then again for me (and maybe for others), it's one thing to read a book (also valuable) and another one to read or engage into discussion.

And one more thing, it takes a great man to acknowledge if wrong (as you did even with me), and another to slip dirt under the carpet (talkin 'bout 1080psf50).
I, for one, dig the difference ;)

Lachrymology
Offline
Joined: Mar 16 2011

The local Panasonic dealer sent me the official response to the 800TVL issue.

It says:

"To measure and fix camera specifications, the “ITE Resolution Chart for High Definition TV” is used. As is well known, tapered lines are used to determine the horizontal resolution. Because the image being shot by the camera is created on pixels of the image sensor (digital pattern), the positional alignment of the detail lines and sensor pixels is quite important. The camera will be positioned as to get the best resolution and measure it.
We also check a waveform monitor and observe the horizontal lines, which clearly separates the lines as four separate waves.
Through performing the above measurement, we are able to confirm that the resolution of the AG-AF100 is 800 TV lines."

There's an image in the doc that appears to show 800 horizontal lines (not sure how to attach it here). I guess the difference between their result and Alan's is that they are not basing it on alias free lines. They also seem to be saying they take the highest reading as the res.

MAGLINK
Offline
Joined: Mar 8 2007

I think as far as the AF101 is concerned people have to treat it as the way panasonic intended it to be marketed as a DSLR replacement but without any of the associated problems of lack of audio and artifacts.

From what I have heard on DVX it makes some very nice pictures and people like it as it has the flexibility to put different lenses on it and make shallow DOF images.

OK it may only resolve 600-800 lines but to compare it to a full broadcast camera or even a 1/3" or 1/2" 3 chip camera is not what it is about as lots of people seem to be finding it very useful. The F3 is now coming out but that is a different camera and priced to do different things, the other sony just released also seems to only resolve around 800 lines too so it will interesting to see what the market takes up as its shallow DOF solution in this price range.

Alan's test was spot on but as we all know not all cars can be a buggati veyron, the AF101 is built to and designed to a cost level and seems to do the job it was intended for.

rone01
Offline
Joined: Feb 9 2009
Gary Nattrass wrote:
I think as far as the AF101 is concerned people have to treat it as the way panasonic intended it to be marketed as a DSLR replacement but without any of the associated problems of lack of audio and artifacts.

From what I have heard on DVX it makes some very nice pictures and people like it as it has the flexibility to put different lenses on it and make shallow DOF images.

OK it may only resolve 600-800 lines but to compare it to a full broadcast camera or even a 1/3" or 1/2" 3 chip camera is not what it is about as lots of people seem to be finding it very useful. The F3 is now coming out but that is a different camera and priced to do different things, the other sony just released also seems to only resolve around 800 lines too so it will interesting to see what the market takes up as its shallow DOF solution in this price range.

Alan's test was spot on but as we all know not all cars can be a buggati veyron, the AF101 is built to and designed to a cost level and seems to do the job it was intended for.

Well said Gary. As I stated earlier - it's only a camera - there are better options but at much greater cost. Folk have got lost on a mission with this camera, elevating and comparing it to everything out there from RED to EX1 - and it's neither. Sits fine with me as someone who at one point struggled with adding a Letus 35mm adapter to a HVX200 with the extra cost, weight and impractical nature of the thing.

The people that will likely purchase it won't give a hoot at its inability to generate loads of alias-free charts.

You get what you pay for.

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

MAGLINK
Offline
Joined: Mar 8 2007

And according to this it's approved for full BBC HD use with the nanoflash: http://soundcloud.com/mm-pubs/bbc-big-sensor-day-af101

rone01
Offline
Joined: Feb 9 2009

Woah, this is a bit of a curve ball isn't it!

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I don't know whether the BBC's approved it or not, but I suspect it would be with significant caveats if it has. The EBU's preparing a formal methodology for categorising cameras according to measurement results, to assign a 0~100 scale to each camera (one of the reasons why I've been so busy for the past few months). So, approval could mean a lot of things, like 'ok for specific shots in specfific programme-genres'. Soon there will be no blanket approval status, which is exactly as it should be.

I like the AF101, it's a decent camera at the time and price. It has problems, which could have been avoided at design time, but which can be lived with by people who know exactly what they're doing.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

drc_online
Offline
Joined: Apr 19 2010

Well Alan it looks like it's not just you who gets mis-quoted all over the Internet.
Since this was me speaking I think I know what I said, and what I said was that, with certain caveats (as you suggest Alan) and for certain types of production (I.e. You have to ask first) it was looking promising.

I am not in a position to grant BBC approval and would not presume anything until it's officially stated on the appropriate BBC commissioning website.

Dave

Shame whoever recorded it mis-spelled my surname too!

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

There's a lot of pillocks out there :)

The Sony PMW-F3 report has gone to Daniel to be put on-line today. I haven't checked to see if it's happened yet.y

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Dugi
Dugi's picture
Offline
Joined: Jul 2 2009
Alan Roberts wrote:
There's a lot of pillocks out there :)

Oh yes Alan. Oh yes!

MAGLINK
Offline
Joined: Mar 8 2007
Dugi wrote:
Oh yes Alan. Oh yes!

And that's the modern noo mediah world we live in where experience seems to be totally de-valued and ignored! :(

Dugi
Dugi's picture
Offline
Joined: Jul 2 2009

Not quite what I meant Gary but, Hey Ho!

MAGLINK
Offline
Joined: Mar 8 2007
Dugi wrote:
Not quite what I meant Gary but, Hey Ho!

It was meant to be tounge in cheek but having just seen a shed 7 programme on BBC 3 where some bimbo called Stacey Dooley tries to solve all of Thailand's political problems I do wonder where meediah is heading these days.

Maybe shallow DOF on unauthorised camera's and codec'ai is the new art form and is so BBC anarchy in the UK? :D

rone01
Offline
Joined: Feb 9 2009

Red EPIC / Scarlet - Panasonic AC 90 -  besq.co.uk

StevenBagley
Offline
Joined: Aug 14 2000

Seems an eminently sensible list except that the HDC950 studio camera isn't listed -- wasn't the BBC's first (modern) HD OB truck fitted with these?

Steve

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

I did comment on this ages ago, but since it was the only available HD system cameras at the time, testing was hardly relevant (but I did actually test it anyway, #Add 12). It seems like an oversight. Not exactly my problem though.

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.