Bolex D16: RAW, 16/Super 16 sensor size, costs less than MkIII

Whether it is the Digital Bolex or not, the first company to market a practical high quality 16mm format digital raw camera system for less than about $6k or so will get my money.
In the mean time, a GH2 is a nice cheap alternative to play with.
 
Okay, I figured out how to demonstrate the difference between true 1920x1080x4:4:4 resolution, and what you'll get from a 1920x1080 Bayer sensor. Which isn't exactly what the DBolex is, it's a 2K sensor, but the same ratios hold.

For those who tuned out (and I don't blame you) the central question was: will a 2K raw Bayer-pattern sensor deliver what is claimed on the specs (i.e., "2K resolution at 4:4:4"). I say it cannot possibly be done, and some folks disagreed, so I left it at "well, when I get my DBolex, I'll test it out and show the differences."

Turns out I don't have to wait. I have a perfectly good simulation here with the GH2. The GH2 has a Bayer-pattern sensor and can shoot stills in raw format. So, with a little cropping, you can indeed get a 1920x1080 raw image off a Bayer sensor using the GH2.

Accordingly, here's what I did -- I framed up a color resolution chart (the famous "Wringer" I'm so fond of) using ETC mode. ETC gives you an exact 1920x1080 frame size, so it's very useful for framing, but NOT for recording! ETC mode doesn't record raw, it records compressed, so I didn't want that. I just used ETC for framing. I then switched back to large (4976-pixel-wide full-sensor 16:9 raw stills) and shot a still. In photoshop I cropped it to the central 1920x1080, giving me an exact simulation (for resolution and color resolution purposes) of what the DBolex would give.

Then, I zoomed in and framed up the chart in full mode and shot it in Large size, raw. I brought that into photoshop and scaled it down to 1920x1080. The reason for doing this is because when starting from such a grossly oversampled source, the net result should be the best we can get in terms of a truly saturated, full-resolution, 1920x1080x4:4:4 image.

I then overlaid the two images on top of each other in a PhotoShop document, so you can click the top layer on and off and compare.

Does a 1920x1080 raw bayer-pattern image deliver the resolution and sharpness we would think we'd get when hearing the specs "1920x1080 raw 4:4:4"? I will let you be the judge.
Overlaid photoshop document here.
 
Barry_Green said:
Okay, I figured out how to demonstrate the difference between true 1920x1080x4:4:4 resolution, and what you'll get from a 1920x1080 Bayer sensor. Which isn't exactly what the DBolex is, it's a 2K sensor, but the same ratios hold.

For those who tuned out (and I don't blame you) the central question was: will a 2K raw Bayer-pattern sensor deliver what is claimed on the specs (i.e., "2K resolution at 4:4:4"). I say it cannot possibly be done, and some folks disagreed, so I left it at "well, when I get my DBolex, I'll test it out and show the differences."

Turns out I don't have to wait. I have a perfectly good simulation here with the GH2. The GH2 has a Bayer-pattern sensor and can shoot stills in raw format. So, with a little cropping, you can indeed get a 1920x1080 raw image off a Bayer sensor using the GH2.

Accordingly, here's what I did -- I framed up a color resolution chart (the famous "Wringer" I'm so fond of) using ETC mode. ETC gives you an exact 1920x1080 frame size, so it's very useful for framing, but NOT for recording! ETC mode doesn't record raw, it records compressed, so I didn't want that. I just used ETC for framing. I then switched back to large (4976-pixel-wide full-sensor 16:9 raw stills) and shot a still. In photoshop I cropped it to the central 1920x1080, giving me an exact simulation (for resolution and color resolution purposes) of what the DBolex would give.

Then, I zoomed in and framed up the chart in full mode and shot it in Large size, raw. I brought that into photoshop and scaled it down to 1920x1080. The reason for doing this is because when starting from such a grossly oversampled source, the net result should be the best we can get in terms of a truly saturated, full-resolution, 1920x1080x4:4:4 image.

I then overlaid the two images on top of each other in a PhotoShop document, so you can click the top layer on and off and compare.

Does a 1920x1080 raw bayer-pattern image deliver the resolution and sharpness we would think we'd get when hearing the specs "1920x1080 raw 4:4:4"? I will let you be the judge.
Overlaid photoshop document here.
Click to expand...

You are a very ANGRY, EVIL, man destroying DREAMS with COLD, HARD, FACTS Barry Green! I think I shall call you "Barry MEAN" from now on!!! :evil:

(Great info BTW) ....I think I will cancel my order, lol! :)
 
4:4:4 for me = I:Don't:Care

seriously, i don't get those who demand such colorspace for small projects which are not effects heavy

great info barry, but shouldn't the point really be: who the heck REALLY, TRULY, JUSTIFYABLY, needs 4:4:4

and before i get a swarm of effects people coming at me with 8:8:8 sized sticks, let me clarify: it's a CHEAP camera, meant for mostly CHEAP productions, NOT big or even medium budget movies.


i always hate myself in the morning after using all caps.....


anyways, im semi interested, if it turns out to be a 3K$ ikonoskope i'll snatch one up
 
if it turns out to be a 3K$ ikonoskope i'll snatch one up
Click to expand...

LOL! You and every other indie film maker on the planet! I'm just hoping that it sees the light of day. (About 100 other people are too at this stage.) Cheers!
 
Hey, like I said, I knew all this and still bought one.

I just dislike misinformation and I'd much rather have it cleared up for folks BEFORE purchase.
 
Thanks for doing the test Barry. The scaled down picture certainly looks sharper and has better colors. Now wouldn't it be nice if Panasonic can just provide a good scaler on the AF200?
 
We'll see how it shakes out. I can definately see the difference and it's a bit more than I'd have hoped. But obviously down scaling a higher resolution image would require processing power within the camera and thus exponentially make this low-cost/simple camera idea more complicated, right? So I'll take the good with the bad for now.
 
Last edited:
There is no question of the benefits of oversampling. Now pull a 1080p 4:2:0 downsampled frame of the same chart from an AVC video file and see how that compares.
 
Barry_Green said:
Hey, like I said, I knew all this and still bought one.

I just dislike misinformation and I'd much rather have it cleared up for folks BEFORE purchase.
Click to expand...

Awww...shucks, Barry. I'm just messing with you. I know you knew. I'm also certain that you'll have mad crazy fun either way. Cheers to you Barry Mean! ;)
 
if anybody can do 1080p 444 its a panasonic ...
they got all the pieces of the puzle...
experience ( hpx3700 do 444 )
they can use next generation sensor from gh3 ....
and they have avc ultra in few flavors ...
but it will be 10000$ + camera....
 
I actually like both images. The downsampled image is really sharp and nice, but the raw crop has that sort of filmic haze to it.
 
vladszp said:
if anybody can do 1080p 444 its a panasonic ...
they got all the pieces of the puzle...
experience ( hpx3700 do 444 )
Click to expand...
Yes, but -- back to the original question -- it does 444 because it's three chips. And three chips is what you need to get 444. Well, either three chips, or a sensor that is oversampled to at least twice horizontal and twice vertical (meaning, you'd need a 4K chip to deliver 4:4:4 2k). Which is what Canon does in the C300.

they can use next generation sensor from gh3 ....
and they have avc ultra in few flavors ...
but it will be 10000$ + camera....
Click to expand...
If the GH3's sensor family were to be developed such that it supports 100% readout continuously at at least 24fps, then - yes, that might indeed seem possible to deliver a 2k 444 out of that system. But I would certainly question whether a low-cost camera like the GH3 would be developed to support a full 18 megapixels (or whatever) at full 24fps continuous readout.

The only camera I know of that could do it right now is the Red or the C300, because they have a true 2x horizontal and 2x vertical oversample at continuous 24fps, so it can provide true 444 color information in a 2k frame. But, of course, the Red can't do it when it's shot in 2k mode! You'd have to shoot 4K in order to have enough color sample information so you could downconvert to 2k and have a full 4:4:4 image.
 
Razz16mm said:
There is no question of the benefits of oversampling. Now pull a 1080p 4:2:0 downsampled frame of the same chart from an AVC video file and see how that compares.
Click to expand...
To what purpose? Again, seems like trying to confuse the issue. The question was: can a Bayer sensor deliver true 4:4:4 at its stated resolution? And the answer is clearly "no it cannot." So what does AVC have to do with that?

Now, if you want to ask an entirely different question, which is: "how would a 1920x1080 AVCHD 4:2:0 file compare to a 1920x1080 raw Bayer file", then that's a fair question and can be answered. But it's an entirely different question.

Simple enough to answer though. I'll just put the GH2 in ETC mode and shoot a 1920x1080 AVCHD file on it.
 
Barry_Green said:
To what purpose? Again, seems like trying to confuse the issue. The question was: can a Bayer sensor deliver true 4:4:4? And the answer is clearly "no it cannot." So what does AVC have to do with that?

Now, if you want to ask an entirely different question, which is: "how would a 1920x1080 AVCHD 4:2:0 file compare to a 1920x1080 raw Bayer file", then that's a fair question and can be answered. But it's an entirely different question.

Simple enough to answer though. I'll just put the GH2 in ETC mode and shoot a 1920x1080 AVCHD file on it.
Click to expand...

Shoot a full frame downscaled version too please. For me the question is not so much whether 2k Bayer raw converted to a 12 bit or 16 bit RGB format can produce a true 4:4:4 but can it resolve color as well or better than typical 1080p 8 bit video codecs at 4:2:0 or 4:2:2. Both discard color information for reduced bandwidth. Your GH2 test is a fair comparison as it removes unknown variables between different cameras for AVC 4:2:0 anyway. If someone could pull a still from an XF100 or XF300 video file for a high grade prosumer video codec that would be interesting too. Canon's 50Mbps 4:2:2 codec looks damn good to me.

PS: Who makes the wringer chart? I could use one of those.
 
Last edited:
Hi Barry have you downloaded and looked at any of the ikonoskop dii sample images? They are DNG files from a bayer pattern ccd imager, in fact, this is the digital bolex in reality really - apart from the lack of 2k and the fact that they are more expensive. I think they even use similar imagers to the one feted to be in the digital bolex. I don't' think they are making any claims to delivering 4:4:4 though
 
there are rumors that gh3 sensor will be global shutter (no need for mechanical one) and if this is true that will be ideal for full readout ...
af200(maybe another name or class) will use tweaked sensor version similar as af100 ...
RAW is simply ALL data from sensor nothing is left out ...
is it 444 or 422 depends of sensor design 3 CCD or one (binning or not)...
 
Last edited:
You must log in or register to reply here.
Back
Top