RED Windowed Sensor Target Area [Archive]

tj williams

03-07-2007, 10:28 AM

Rob, here goes:

Stuart/Graeme I would like a clarification here concerning the recording area of the windowed sensor. Several posters who seem to know the camera well are inferring in several different threads that the size of the windowed sensor is variable ie that you could use lenses covering a smaller area than 2KRAW if shooting RGB720 etc.

Super 16mm exposed area is 12.35 by 7.5 mm The Abakus adapter for B4 lenses converts the lens to cover this dimension for super 16mm use with Pl mount. It also of course converts a lens intended for a prism and three targets into a single sensor/film plane lens.

Let me know if the below is correct.

I had concluded that the RED windowed sensor was recording a target this size(Super 16 recorded area: 12.35 by 7.5 mm) regardless of whether the image being recorded was 2K RAW or direct to 1080P or 720P Is this true?

I'm looking at buying a converter for our HD lens so have a need to know this!

Andrew M.

03-07-2007, 10:56 AM

Two windowed sensor areas are currently allowed -
4K - Academy 35 mm
2K - Super 16 mm / B4 using adaptor
4.5K (not a window of course) - Super 35 mm

Note 1080p and 720p recordings are designed to have the same FOV, due to the design of the internal scaling engine. This means that 1080p, 1080i and 720p footage will all have the the same optical characteristics...

See:
http://www.reduser.net/forum/showthread.php?t=406
http://www.reduser.net/forum/showthread.php?p=7689&highlight=windowed#post7689

Andrew M.

03-07-2007, 11:08 AM

TJ read this link about chromatical abberation. (second page)
I was also looking to use HD lenses.
I am not sure now if it is a good idea.

http://www.cinematography.net/Pages%20DW/HDLensesOnFilmCameras.htm
http://www.edmundoptics.com/techsupport/displayarticle.cfm?articleid=261&search=1/

Brook Willard

03-07-2007, 01:37 PM

Academy 35mm = 22mm by 16mm
RED 35mm/4K = 22.2mm x 12.6mm

4K is not Academy 35mm.

tj williams

03-07-2007, 01:52 PM

Andrew,

Academy is 1/37 to 1 without a 1:85 mask I believe the 4K area is 1:85 height width ratio, thus not academy? (brook you got in there before me TJ)

You are def. a better searcher than I. I was not able to find the threads you site. Glad to confirm that the 2K 1080 720 windows are all the same as S16!!

The Prism effect and the blue setback are of course well known. I talked to Les Zelig who owns ZGC and who sells the Abakus adapters for S16 and 35mm from B4.

He says this:

These Adapters have been tested and there are no fringing/focus artifacts. If you have a problem with this adapter then you will have the same problem with any S16 lens according to Les.

He is concerned that some folks might use the adapters to put not so good video lenses on the RED and harm picture quality. HD lenses generally are going to be good enough for this application according to Les.

Les says that NFL Sports uses these adapters for filming football in S16 with wonderful results they are very pleased. He says the adapters, are all metal and so rugged that the lens B4 will fail way before the adapter. He was holding one in his hand as he said this! Also Les is personally quite a RED fan and told me that he thought the sellers of very expensive HD and Electronic Cine. cameras at NAB were in for a nasty shock!

The S16 Adapter costs you about 1/3 stop on your lens as the magnification factor from 2/3 to S16 is 1.32 The larger magnification factor in converting a B4 lens to 35mm is greater than I had thought about 2.5 stops of light loss. Les feels that these (35mm adapters) were only made for special applications and that the S16 lens converter is the way to go.

Ken Willinger

03-07-2007, 02:13 PM

TJ, I thought RED was going to make a B4 adapter available as part of the RED system. Will the converter you are speaking of be needed as well as the RED B4 adapter?

tj williams

03-07-2007, 02:46 PM

Getken I'm unaware of any lens mount made by RED except for Pl and Nikon. Maybe a RED team could chime in here. The Abakus mounts your B4 lens directly to the PL adapter which comes with the RED.

There are issues with B4 lenses (video HD lenses) which must be addressed because they are set up for 2/3" chips which are smaller than S16 and to shoot thru a prism at three chips which are not on the same plane. It's not just a B4/Pl adapter, it also works optically to adapt to S16 size imager. and single sensor.

Andrew M.

03-07-2007, 02:55 PM

TJ, I understand that these adapters/lenses combinations are good enough for final output for HD TV presentation.
If you shoot material that will end up next week on HDTV then all this chromatic collimation/focus problem will not be really visible.
Remember HDTV max resolution is 2Mpixels but what really gets home to the TV viewer is maybe 1MP. One day, cable and over the air transmission will use 100 or maybe 300 Mbits/sec bandwidth and then we will actually see all 2MP at home.
So if your footage is for the near and future (5 Years) presentation then we better look in to this chromatic collimation more carefully, looks like it is serious problem as soon as you present your material in 4:4:4 instead of 4:2:2 that has 25%!! of full possible chroma resolution only.

I remember the discussion about how many pixel equivalent you have on 35mm film.
Someone mentioned 6MP RGB, if you look at film as a chemical particles. However the film is never flat as CMOS chip is. To keep perfect focus on the whole surface of the frame when you record on the film and then when you project it back on screen, is almost impossible. So most people agreed that 2MP equivalent is probably the max what the best film/camera/projector full chain can deliver. Now we will be using 4MP CMOS chip, perfectly flat, no film curvature or imperfections, perfect flat surface.
( I know you will tell me you will be using only 2MP in windowed 2K.)
Well it is the same 25 micron pixel servicing 2K or 4K or 4.5K formats so the chroma collimation/focus error will be actually more magnified on the 60’ projection screen from 2K windowed then from 4K or 4.5K full CMOS shots. So when using HD lenses we are loosing twice, once because we are using smaller window that have to be magnified to the full screen size, whatever it will be, second we are loosing on the precision of the focus and other aberrations the lenses deliver.
One thing for sure, bad 35mm lenses will not give you better picture then top of the line S16 or HD one. I wish that Abacus would come up with the adapter that corrects chroma focus problem. I don't think the current version corrects chroma problem, otherwise it would be trumpeted all over on abacus website. But if you can get good chroma corrected and whatever??? corrected S35 lenses and not blow up your budget then it will be a dream.

Scott Webster

03-07-2007, 04:25 PM

tj williams

03-07-2007, 04:30 PM

Andrew
very interesting stuff. thanks for the posting mostly when shooting in a windowed mode I will be shooting HD for current consumption. As you say if this is so it is only a mid term issur.

The CMl link above also contains the arguments of guy "A" who had shot several features with an HD lens on S16camera.. apparantly without any optical correction at all. so the difference in S16 blowup, is fairly subtle?

When I look at current HDtv and note the percentage of shows all the nets allow to be up-converted fm Sd, I'm not so sure that the future will have this much more rez. But hey it is the future.... so who knows.

Since Les is a pretty good guy, and very honest from my own experience, and since he seems pretty certain that the adapter is good for any 2K use, could you go into a little more detail as to where you have gotten the info that the adapter is limited in this way. On the abakus web site not much technical specifications there. So can you go into a little more detail as to how you know this.... Have you used one of these or tested one, or know someone who has? I don't mean to cast any aspersion, but a lot of stuff is presented here as fact which is opinion.

1. it is obviously solid and holds the lens firmly on the plane
2. It contains optical correction for the size of the S16 imager
3. You are saying that this optical correction does not correct

all this chromatic collimation/focus problem
address these problems of HD lenses.

Brook Willard

03-07-2007, 05:23 PM

There will be a B4 adapter for RED. It will use a 2/3" sensor area.

tj williams

03-07-2007, 06:15 PM

Brook so you are saying that the RED developers have decided to put in an additional size window (2/3") on the sensor and make a B4 adapter themselves.
is this a new post from RED developers or just hopeful?

Brook Willard

03-07-2007, 06:23 PM

This has been the case for a while. It's on www.red.com in the format options and a mount is planned.

tj williams

03-07-2007, 07:10 PM

HI Brook.

Andy refers to this post:

on 1/26/2007 Stuart says:

Default Windows
Two windowed sensor areas are currently allowed -

4K - Academy 35 mm

2K - Super 16 mm / B4 using adaptor (this means an adapter to S16 size window?)

4.5K (not a window of course) - Super 35 mm

[fixed typo ~Brook] (Looks like you fixed his english?)
Brook can you explain this? It appears to me that Stuart who is certainly in the know is saying that there are two windows only?
__________________
Workflow Wizard

Ken Willinger

03-07-2007, 07:16 PM

TJ, I don't know which thread it was but fairly recently there was a thread in which JarRED had been testing the camera with a 2/3" lens on a B4 mount which apparently was very succesful. That's why I'm under the impression that there will be a RED B4 mount. And it seems Brook is confirming this.

Scott Webster

03-07-2007, 07:35 PM

No, Jarred tested Spike with the Abakus mount, not a Red version. No detailed reports were forthcoming on how well this did or didn't work and what issues (if any) presented themselves.

That abakus is what we used.. its an optical coverter, it uses (i believe) 2 elements and by guesstimation i would say it lost just under a stop of light.

The Lens was a Fuji HA13x45 1.8 lens. The zoom range is 4.5-59mm.

Scott Webster

03-07-2007, 07:47 PM

Original B4 test thread:

http://www.reduser.net/forum/showthread.php?t=392&highlight=abakus

Brook Willard

03-07-2007, 08:39 PM

Cropping to 2/3" was originally planned. Judging by Stuart's post, I suppose any such cropping would now have to take place in post.

Ken Willinger

03-08-2007, 04:12 AM

Thanks for the clarification. I look forward to the results then as I plan to use B4 lenses.

Jarred Land

03-08-2007, 06:50 AM

Things can and most likely will change, but at this point, we are very confident with the abakus B4 solution

Andrew M.

03-08-2007, 07:57 AM

TJ I am not the expert but I am fast finder/reader.
See the link and the last page of the document from 2002 the references section last item.

http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/DigiPrime-Artikel-KMu_SMPTE2002/$File/SMPTE2002_DigiPrime.pdf

Very good reading.
Now as it comes to the Abacus B4 mount, we should ask the question straight by e-mail to the Abacus customer service. From different sources we know that Abacus has 2 glass elements in it. I can’t imagine how you can correct chromatic focus mismatch using only two elements. IMHO, FOV correction yes but anything else no.
Now, good HD lenses will be almost impossible to use without correction of chromatic focus on the perfectly flat CMOS surface. You will buy 25K HD lenses just to perform like 3K one?

tj williams

03-08-2007, 08:06 AM

Here is the Abakus response to the problems with their unit presented by Andrew: That second was an interesting post Andrew. (we think alike by the time I read that I'd already Emailed Abakus) So if we accept what is stated below then Digi primes are also available to RED owners who have an Abakus adapter. Getkin, Looks like good news fm Abakus that will enable those of us who already have good B4 lenses to use them. Brook guess I don't get it if Jarred says Abakus is the solution and Abakus says they have overcome the above problems, then how will you trim it in post? What will you trim in post? where will you B4 mount the RED?

Hello TJ,

Thank you for your email, I will try to answer some of your questions.

Video lenses are designed to compensate for the beamsplitter block in the camera that splits the light into red, green and blue.
When we designed our converter we re-compensated for this so that only air should be in the back focus of the lens. This is the position on super-16 and for single chip detectors.
The converter therefore DOES optically compensate for the longitudinal colour (and the spherical aberration) that would otherwise occur. It does not just magnify the image. Our website is obviously too conservative in not emphasising the point.
We did not compensate for any film curvature in the design (perhaps we should have?) since it would vary with different cameras

The design of our converter assumes that the video lens is perfectly designed and manufactured for the Sony HD block (we managed to persuade them to give us the details).
Whilst most HD lenses are fairly good they are obviously not perfect. Since the converter magnifies the image 1.32 times then the errors are also magnified. It certainly shows up any flaws in the video lens.

As far as the discussion on the effective resolution of film versus electronic detector, much has been written. Most subjective test indicate that film resolves better than it should. I suspect that this is due to the averaging effects of having a new detector every 24th of a second. Although the grain size of the film is larger than the pixel spacing on a CMOS detector and the film does not locate in exactly the same position for each frame, the human brain is very good at averaging these out and the perceived picture can be much better than achieved on present day HD cameras. This is certainly true for 35 mm and to a lesser extent for 16 mm.

Please contact me if you have any further technical questions.

regards

ken

Abakus Ltd. Registered in England, No. 2595892. - Registered Office Grange Farm, Bourne Road, Carlby, Stamford, PE9 4LU, UK

Andrew M.

03-08-2007, 08:25 AM

TJ It is great news, I just sent the e-mail myself to Abakus, asking them how many glass elements they have in the B4 adapter, than we do some science on this info.
I would love to use B4 adapter that stretches the picture to full S35 size, maybe we can suggest to Abacus to manufacture one. The current one I believe is only stretching it to S16 size. Could you ask them this question, please? Here goes another stop of light:-)

Andrew M.

03-08-2007, 08:42 AM

As far as the discussion on the effective resolution of film versus electronic detector, much has been written. Most subjective test indicate that film resolves better than it should. I suspect that this is due to the averaging effects of having a new detector every 24th of a second. Although the grain size of the film is larger than the pixel spacing on a CMOS detector and the film does not locate in exactly the same position for each frame, the human brain is very good at averaging these out and the perceived picture can be much better than achieved on present day HD cameras. This is certainly true for 35 mm and to a lesser extent for 16 mm.

I will not agree with this but 4K will kill the 35mm film once and for all:-)
He forgets that there is other part of making the movie, it is displaying it.
The same lenses and film curvature and other errors are made three times.
Once when you acquire the data, second when you process/copy/distribute, third, when you display the data (film)

In the digital world, the errors are acquired once when you record, then you mask all imperfections in the electronic massaging/editing improving the data and finally there is no errors add by watching it on large HDTV screen.
I agree, CbaleTVs companies screw the signal beyond the recognition trying to deliver it to your home but IPTV will take care of it very soon. Digital cinema will not degrade the data as much as the film projector does, if not actually improving it.

tj williams

03-08-2007, 10:46 AM

Andrew of course there are many opinions about quality and the future. The people at the adapter company are more film centric as that has been the primary use of their product. The opinion you have comes from being more attuned to digital tech. The important thing to me is that the adapter does address the problems of HD glass since I have a very expensive piece of HD glass the opportunity to use this is phenom. for me! You do not want a 35mm size converter. They make one. It will lose 2.5 stops of LIGHT!!!! This not because of them this is the physics of any optical device!

Andrew M.

03-08-2007, 04:43 PM

TJ, more I read and more I talk with people, more I see that we just have to try it.
Lens manufactures do not want to disclose details of the glass structure and color responses for each piece of glass in it. So it is difficult to say what will work with what.
As to magnifying 2.5 times to get B4 lenses to illuminate whole S35 I agree that we have to loose the light, but don’t we gain it back by downresing it back to 2K? Then 4 pixels are combined in to one so the signal to noise ratio goes up by 12dB?
Well the Abakus adapter is not that expensive, so if you are ready buy new one, and try it. The good thing is that you know your lenses very well so you will be able to tell right away if something is not as should be. However you will be trying it on the RED that is completely unknown to you.
What kind of HD lenses do you have now?
If your experiment will not work out good then I will rent few HD lenses and I will try it again. What I have heard, some lenses design is not lending itself for any glass between the sensor and the lenses, some lenses is very difficult to correct for chromatic focus and so on, nothing scientific and solid, just the talk. Yes, I am more science guy so numbers and data is more appealing to me but I have to take subjective opinion of practical experienced people very seriously.

Michael Hastings

03-09-2007, 08:05 AM

I found the comment about film providing a "new detector" every frame very interesting. For the video guys what he is saying is that the fact that the film grains are in a slightly different position in every frame it works sort of like pixel offset on a 3 chip camera where the chips are aligned so that the pixels on one chip are aligned to the spaces of the other chips - dramatically increasing resolution of the final image - it is not double - a sony engineer explained that it is a square root factor so about a 1.4 increase. In the case of movies it is temporal - the offset occurs on the next frame but persistence of vision and the brain figure it out. It also reminds me of a software package I have seen at NAB that takes multiple frames of a video to create a much sharper still image. It doesn't work if the camera is solid on a tripod, it only works if there is some CAMERA movement so that they get more information about the scene to process. Again it doesn't work on the moving subject - it works on the change in movement of the camera (and therefore the sensor) And it definitely worked - the examples I saw took a .35 megapixel standard def video image and made what looked like a 1 or 2 megapixel still.

As far as the discussion on the effective resolution of film versus electronic detector, much has been written. Most subjective test indicate that film resolves better than it should. I suspect that this is due to the averaging effects of having a new detector every 24th of a second. Although the grain size of the film is larger than the pixel spacing on a CMOS detector and the film does not locate in exactly the same position for each frame, the human brain is very good at averaging these out and the perceived picture can be much better than achieved on present day HD cameras. This is certainly true for 35 mm and to a lesser extent for 16 mm.

Please contact me if you have any further technical questions.

regards

ken

Abakus Ltd. Registered in England, No. 2595892. - Registered Office Grange Farm, Bourne Road, Carlby, Stamford, PE9 4LU, UK

Andrew M.

03-09-2007, 08:27 AM

I found the comment about film providing a "new detector" every frame very interesting. For the video guys what he is saying is that the fact that the film grains are in a slightly different position in every frame it works sort of like pixel offset on a 3 chip camera where the chips are aligned so that the pixels on one chip are aligned to the spaces of the other chips - dramatically increasing resolution of the final image - it is not double - a sony engineer explained that it is a square root factor so about a 1.4 increase. In the case of movies it is temporal - the offset occurs on the next frame but persistence of vision and the brain figure it out. It also reminds me of a software package I have seen at NAB that takes multiple frames of a video to create a much sharper still image. It doesn't work if the camera is solid on a tripod, it only works if there is some CAMERA movement so that they get more information about the scene to process. Again it doesn't work on the moving subject - it works on the change in movement of the camera (and therefore the sensor) And it definitely worked - the examples I saw took a .35 megapixel standard def video image and made what looked like a 1 or 2 megapixel still.

With the film though, I will not agree that brain is somehow good in averaging the picture to the benefit of higher resolution perception at 24fps.
If you look how eye is build and works, plus the persistence time of the eye sensors, picture shift averaging on the retina, will just produce more out of focus perception at low frame rate. More jitter there is at 24fps more out of focus perception of the picture will be. For 120fps (minimum 60fps) though, if film will offsets a bit up and down and left and right, it may work. It is like having not only horizontally interlaced image but vertically as well.
For the camera movement and software that squeezes out more resolution from it, I agree 100%

Stephen Williams

03-09-2007, 08:46 AM

They make one. It will lose 2.5 stops of LIGHT!!!! This not because of them this is the physics of any optical device!

Hi TJ,

Cooke made a 16mm version of the 20-100 T3.1, I think it ended up T1.5 or 1.7.

Stephen

Andrew M.

03-09-2007, 03:27 PM

Sent two e-mails to Abakus and no answer.
They don’t want even to confirm if it has two glass element in it.
Looks like there is a lot of misinformation out there about lenses.
Well everybody wants to sell its glass but only very few know how to make it right.

Andrew M.

03-09-2007, 03:40 PM

TJ got the answer from Abakus, looks gooood!
Straight from the source.
====================================
Hello, I have one question:
How many glass elements Abakus 260 Converter contains?
------------------------------------
Hello Andrew,
6 elements in 3 groups.
regards

tj williams

03-09-2007, 10:06 PM

I am more science guy so numbers and data is more appalling to me
I hope you mean appealing? I'm the opposite as numbers and data are appalling to me!

I totally don't get the counting glass elements etc. either this thing works or it doesn't and two pretty credible folks say it does?

Stephen This is not a lens adapter what your are talking about is using parts of a lens to manufacture a complete lens. I already have an HD lens and I just can't affore to have Cooke redesign and rebuild it. The deal is the larger you spread the image ie 2/3 to S 16 or 2/3 to 35mm size then in a direct relationship you will loose light. This is straightforward so about a 1/3 stop loss for S16 and a 2.5Stop loss for 35. Cooke or no one else can change that!

Stephen Williams

03-09-2007, 11:33 PM

Stephen This is not a lens adapter what your are talking about is using parts of a lens to manufacture a complete lens.

Hi TJ,

Cooke only changed the very back of the lens, the magnification was smaller giving a brighter image. A teleconverter in reverse, not that different in principle from the adapter you are talking about IMHO.

Stephen

Andrew M.

03-10-2007, 06:21 AM

I totally don't get the counting glass elements etc. either this thing works or it doesn't and two pretty credible folks say it does?

Yes, experience and practical knowledge is very important but thinking out of the box is also beneficial for the progress of science.

Abakus is magnifying image 1.32 times to stretch it to S16. To streach it to S35 requires magnification factor to be 2.4 or so. Yes all errors of the lenses will be magnified and we loose the light but as I mentioned before, what we loose we gain back by downresing it and it is easier to correct aberrations coming down from 4K to 2K then correcting it on the pixel per pixel basis.
Also, if Abakus will do good job on the adapter, without introducing too much of its own distortion, then good lenses versus bad ones will show right away. What I am reading here there are very few B4 ENG lenses out there of good quality.
Maybe we could pick the Graeme’s brain here on this subject:
To magnify the B4 lenses 2/3 image size to S35 and then shrink it in post back to 2K/1080 or just leave it at 2K. (you can shrink it also directly on the camera/CMOS format converter)

IMHO you can correct a lot of imperfections combining 4 pixels in to one, where once you record in one to one pixel ratio with small imperfections, then it will stay there and it is not much you can do about it.

As to the size of CMOS sensors of cameras in the future, the S35 or even 135 will be the way to go, since technology is improving and bigger the sensor is the better S/N (dynamic range) it gets. For cinematography though, I think we will stay on S35 for quite some time. Too many good lenses and accessories are out there for this format and all experience out there is based on it. Broadcast industry will move slowly to the bigger sensor size and they are the quantity users of the cameras out there so this is the force that can change a lot in this market.