23 stops of range?

[Michael.B]

I saw this at the recent HPA retreat. Thomson was showing an image from the GV Infinity claiming to have 23 stops of range. Apparently the light in the images, a 500 watt halogen, is the sole light source (maybe some reflectors in the room). Found the images on cinematography.com and thought people might be interested in hearing about it here.

http://www.reduser.net/forum/uploaded/15111_1235799734.jpg
http://www.reduser.net/forum/uploaded/15111_1235799765.jpg

[Robert Berger]

I saw this at the recent HPA retreat. Thomson was showing an image from the GV Infinity claiming to have 23 stops of range. Apparently the light in the images, a 500 watt halogen, is the sole light source (maybe some reflectors in the room). Found the images on cinematography.com and thought people might be interested in hearing about it here.

http://www.reduser.net/forum/uploaded/15111_1235799734.jpg
http://www.reduser.net/forum/uploaded/15111_1235799765.jpg

Do you see the hard shadows by the botles?
regards Robert

[alex trettenero]

still, very impressive.

[Michael.B]

Do you see the hard shadows by the botles?
regards Robert

yeah there's hard shadows on everything. not sure if that means secondary light source or reflected light.

[Brandon Fraley]

i dont buy that that's the only source.

[Roberto B]

typo.. it's probably 13 instead.. :)

[Herring]

If it's true it's interesting, but this picture proves nothing.
There is clearly a lightsource lighting the scene. Just make it a big one and you will be able to see the filament with any camera.

Joost

[Jeff Kilgroe]

Definitely light coming in from somewhere else and I don't believe, with that shadow, that it's reflected from that 500W halogen. I''ll believe 13 stops, not 23 stops. It's going to take a lot more than a shitty still of cereal boxes, wine bottles and a cheap work light to convince me of 23 stops.

[J. Eric Camp]

Well right off the bat:

The hard shadows are not from a bounced light source. The shadows would be softer. Further more, you will notice a tight specular highlight reflected in the bottles on the left as well as the two on the right just above the halogen. This makes it clear that there is at least one source pointed at the scene. The bottles on the right have multiple highlights which could either be refraction or revealing more sources. Either way I can confirm one source pointed at the scene. The harshness of the shadows and the shape and style of the highlights tell us that it is a directional lamp with a lens.

Past all of that, the image is soft.

I'll be interested in seeing anything else they have to offer. I remember hearing about RED when it was nothing more than an idea.

However 23 stops... nothing in this image shows me that.

[Obin Olson]

bull on this one! you can't have the color card showing the same value as the inside of the light reflector!

[Bill Anderson]

I think your misunderstanding the poster's info- he said the scene is lit by a 500 watt
source - that would be excluding the light source directed toward camera.
Shadows and highlights are consistent with this. The off camera light isn't that close to the subject- look at the light ratio between objects camera right and left. Even then there are too many unknowns. We need a little more data.
It isn't soft- look at the grid pattern on the light detail image. Besides, if this test is for latitude, it's not ideally lit and set up for critical focus assessment. There will be some plane of critical focus but it's hardly worth searching for. Not too shabby, though.

Obin, my black and white stills work using staining and compensating developers tells me you can have the inside of the reflector contained to the extent of showing the nearby card at these values- and the card isn't the same value- and we don't know what light has been pumped onto it. I'm saying this is not impossible, but...

[Finner]

Not sure why so many are hammering down on this. For a pure technical dynamic range indication this appears to be incredible. It will be good to see properly lit images from this camera.

For all I know about this camera its images could look like a crap sandwich with a side order of crap. But if that still has not been doctored then it would definately have an unbelievable amount of dynamic range.

[Bill Anderson]

Totally agree, Finner

[Herring]

Sorry Bill, I disagree on one thing.
Michael writes: "Apparently the light in the images, a 500 watt halogen, is the sole light source". I can only conclude he means the lightsource is in the picture.

Finner - You are right we should be open and curious about new developments. Only this picture - regrettably - does not prove "an unbelievable amount of dynamic range", because it's not clear how it is lit.

Just a simple outside shot in de the sun with lot's of shadows and contrast would tell me a lot more.

So I would love to see more footage from this camera.

Cheers,
Joost

[Dan Hudgins]

With the light pointed into the lens you get lens flare that lifts the shadow areas, in fact if you put bias lights into your RED ONE you could get a A+B=C curve that expands the dynamic range, but would give too much noise in the dark areas since the sensor area is too small.

Shooting with normal lenses, the dynamic range is limited by the lens flare since lens flare can reduce the bit deltas for detail in the darker areas. Lens flare can also pull dark areas up out of the compression and sensor large steps at the bottom of the A2D, so a little can help and too much can hurt...

If people start trying to get clean 23stops from a single lens, they will need fewer element groups and better coating. It should be possable to make a single group cemented multi-element lens with just two air surfaces that would have better flare values, the two surfaces being EBC coated etc.

Anyway, you should be able to get more bits now using two synced RED ONE in an HDR rig if you need to shoot without lighting control.

[Brandon Fraley]

Not sure why so many are hammering down on this. For a pure technical dynamic range indication this appears to be incredible. It will be good to see properly lit images from this camera.

For all I know about this camera its images could look like a crap sandwich with a side order of crap. But if that still has not been doctored then it would definately have an unbelievable amount of dynamic range.

I disagree. If you use multiple lights, as we all agree the original photographer did, you could recreate this image with a RED ONE, or hell, with enough finesse you could get a DVX100 to do it (res excluded)

Being able to see into the shadows and still see the filament would indeed be incredible, but if you add enough light to the rest of the scene so that you can expose for the filament, it's not that big of an accomplishment. right?

[J. Eric Camp]

I am not against anything. I am simply of the mind set that this photo with limited to no support information gives me nothing to get excited about.

Show me clear evidence and my mind set is a different story. I was the same towards the RED ONE when it was first announced.

[Michael.B]

To clarify, I believe they said the light in the image was the only source, but I could be mistaken. There are some mysteries here, but I'm not sure that Thompson would come out with an outright hoax at a high level event like the HPA.

[GlennChan]

To me, it looks like classic HDR.

Take *multiple* exposures.
Merge them with HDR software... which will also do tonemapping for you.
Now you have a HDR image. Any camera can do this*. (*Provided that it is sensitive enough.)

So I'm not sure what this image is really supposed to show, unless you want to use a video camera on only non-moving subjects.

1b- OR... they are doing temporal noise reduction tricks. But again... would you want to use the Infinity on non-moving subjects?

2- To get 23 stops from a camera in a single exposure, you'd probably need a 23-bit A/D...

3- I could very well be missing some context... which I think we are missing.

[Dan Hudgins]

2- To get 23 stops from a camera in a single exposure, you'd probably need a 23-bit A/D...

I don't think that at the pixel frequencies required for Digital Cinema the pre-amps in/on the sensor could have a slew rate that would allow 23bits of accurate data. In a Bayer sensor if you have four pre-amps you can reduce the color de-saturation but you still get low pass and overshoot from the pre-amps so you would not be getting true 23bits for every pixel on all subjects?

[Josh Negrin]

am I taking crazy pills or did that image look like crap? Also, I hear with more stops of range you can tell better stories!

[Dan Hudgins]

am I taking crazy pills or did that image look like crap? Also, I hear with more stops of range you can tell better stories!

Cinematic footage that looks good used lighting to bring the tonal range within the limits of the display, about 9 bits.

If you have a 23 bit image in linear mapped linear or log to a 9bit display range it will look very flat and "soft".

While there are some cases in uncontrolled lighting where extra stops can help get a less clipped film like look, for the most part 12bits is all you need if you light to control the brightness range of the subject to hold it within the center 10bits.

Rent "How Green Was My Valley (1941)" if want to see how cinematic lighting is done, look closley as where the light is comming from and how many fixtures are being used even on simple shots...

[Anthony Gratl]

if that's the only source, than that light is being bounced onto a mirror

[Pawel Achtel]

I actually find the image incredible. It certainly has very high DR that would be impossible to achieve using traditional cameras.

[Brandon Fraley]

I actually find the image incredible. It certainly has very high DR that would be impossible to achieve using traditional cameras.

why do you say this? do you mean this isn't possible with a red one? I think it would be easy if you expose for the filament, use a very bright key up and right of camera, and some fill for the shadows. am I wrong?

[Finner]

why do you say this? do you mean this isn't possible with a red one? I think it would be easy if you expose for the filament, use a very bright key up and right of camera, and some fill for the shadows. am I wrong?

If you think that they are liars then you may be correct. If you trust that they are not then you are wrong.

I mentioned earlier that I feel the amount of dynamic shown is incredible. If it is truely only lite with the 1 open face 500 watt light and silver bounce or mirrors the it has fantastic dr. I don't know anything about the about the camera or the people that are stating that it is only lite with the 500 watt in the still. Because I do not know them I feel I have no right to call them liars and trust for now that the information given is correct. It would definately call for future due dilligence to make sure that what is presented is fact but for now I see no reason to call it bull sh*t.

[Jeff Coatney]

This is fun, but I can't buy into the single source theory for four reasons.

1. The specular highlight on the wine bottles is clipping while the filament on the lamp is not. How can this be?

2. Also the reflections in the wine bottle and on the plastic bottles in front of Ernie's feet clearly show a circular secondary source, besides the rectangular halogen source in the picture, that is aimed down at the set at approximately 37 degrees to the right of camera, producing the hard shadow.

3. While the reflection in the wine bottle indicates a single source light, the plastic bottles reflect three separate hot spots, which could indicate that a single light was moved twice for separate exposures or a tight grouping of three sources for a single exposure.

4. If the light from the halogen is being bounced or re-directed by a mirror, then why do we not see any of the apparatus reflected in the glass and plastic? And if the light is redirected, how can the reflected highlight on the bottle clip, while the filament in the halogen lamp does not?

[Brandon Fraley]

This is fun, but I can't buy into the single source theory for four reasons.

1. The specular highlight on the wine bottles is clipping while the filament on the lamp is not. How can this be?

2. Also the reflections in the wine bottle and on the plastic bottles in front of Ernie's feet clearly show a circular secondary source, besides the rectangular halogen source in the picture, that is aimed down at the set at approximately 37 degrees to the right of camera, producing the hard shadow.

3. While the reflection in the wine bottle indicates a single source light, the plastic bottles reflect three separate hot spots, which could indicate that a single light was moved twice for separate exposures or a tight grouping of three sources for a single exposure.

4. If the light from the halogen is being bounced or re-directed by a mirror, then why do we not see any of the apparatus reflected in the glass and plastic? And if the light is redirected, how can the reflected highlight on the bottle clip, while the filament in the halogen lamp does not?

agreed, seems open-shut to me.

[Jeff Kilgroe]

I actually find the image incredible. It certainly has very high DR that would be impossible to achieve using traditional cameras.

I'm pretty sure I could duplicate that image with the RED One or the F23 or my Nikon... But that's beside the point. We don't know the full details of their setup or any way to confirm anything about this image.

At this point, I'm highly skeptical of the 23 stops, but I want to believe because that sort of range would just be amazing. I was a little down on the image earlier in the thread because it's a crap image that tells us absolutely nothing. As Evolve said, if that halogen lamp is indeed the only light source, then they must be bouncing it off a mirror to get the sort of reflections and light streaking they're getting. But I'm still not convinced it's the only light based with the shadows that are present. If it is in fact the only light, then the DR is very impressive. I don't think we can tell if it's 23 stops of impressiveness from this image, but still impressive. If they happen to be fudging the truth and using a fill light in addition to the halogen, then it's all bogus.

[Pawel Achtel]

I'm pretty sure I could duplicate that image with the RED One or the F23 or my Nikon... But that's beside the point. We don't know the full details of their setup or any way to confirm anything about this image.

At this point, I'm highly skeptical of the 23 stops, but I want to believe because that sort of range would just be amazing. I was a little down on the image earlier in the thread because it's a crap image that tells us absolutely nothing. As Evolve said, if that halogen lamp is indeed the only light source, then they must be bouncing it off a mirror to get the sort of reflections and light streaking they're getting. But I'm still not convinced it's the only light based with the shadows that are present. If it is in fact the only light, then the DR is very impressive. I don't think we can tell if it's 23 stops of impressiveness from this image, but still impressive. If they happen to be fudging the truth and using a fill light in addition to the halogen, then it's all bogus.

Well, I am not sure it is actually 23 stops - maybe, maybe less, but it sure has considerably more DR than any camera I have seen. Assuming it is not just an HDR composite and there are no motion artefacts, I can certainly see some good use of such camera. Just point it on a scene that needs high dynamic range to look good. The example provided in not one of them, but does seem to demonstrate that something is different here. Interesting.

[Dan Hudgins]

Why can't they take a photo of two light meters, one with a 5K spot pointed at it, and another in a black panted box to the other side of the frame, show us that the incident dome reading is 18 stops different, they would need a few extra stops so that the white dome on the hot meter would not be burned out, and the black case would not be blank in the dark box meter?

Why can't they take a back-lit shot of a door way to bright sun so that we can see that the outdoors look good as well as in unlit indoors?

Anyway, HDR or 14bits is the way things will go, maybe 16bits or more in 5years for some DSLR cameras etc.

Right now, few if any 12bit cameras give 12 accurate bits for all three colors, the noise can be much higher so you only get maybe 6 true bits in each pixel and they average the primary colors into luma and low passed chroma to get a lower noise than the RGB would show, and even then the Blue and sometime the Red have more noise than the Green.

Once they pass 16bits of accurate data the future use of film would be for what?

[Anthony Gratl]

4. If the light from the halogen is being bounced or re-directed by a mirror, then why do we not see any of the apparatus reflected in the glass and plastic? And if the light is redirected, how can the reflected highlight on the bottle clip, while the filament in the halogen lamp does not?

hmmm....well not sure why you think you'd see the apparatus (i assume you mean the physical light itself yes?) reflected in the glass and onto the bottles. The clipped highlight on the bottle could be from catching both peripheral light from the light itself as well as the reflected light.
Also, as a hypothesis, i'd imagine that the bottle has difference absorbent qualities than the filament. Perhaps the bottle is reflecting the light back onto the mirror which is then reflecting back yet again, and so on and so on, thereby causing the clip.
I dunno, I don't see a reason to fudge the test if they're trying to prove that they have great dynamic range. Kinda defeats the purpose no?

[Bill Anderson]

Herring, you might be right about the light pointing at the camera being THE light source, but the poster is most vague about 'reflectors' and then goes on to say he's NOT SURE if a secondary light source was used. If the light in shot is the only light source, THAT would be an achievement. We need more info.

I'm not sure if it's a perfect analogy (given the post differences) but in stills work there's a point when long tonal range gets too much for most tastes, and ironically is often countered with harder- higher contrast- paper grades. I'm sure you've all seen platinum prints from long scale negs: they can be quite beautiful for the occasional interpretation but would be too bland for most scenarios. I'm sure digital isn't quite there yet but there will likely be a point when added tonal range will mean a fair amount of post compensating. Still, it's better to have it and not need it.. .
JN Films- no, the image doesn't look like crap if you're looking for tonal range potential; if you're looking at it as a cereal commercial then it's crap- And having fewer stops won't help your stories any more than being restricted to 1/3 inch sensors will. This is simply about giving us choices- like red's incredible range of sensor development.
Brandon this scenario could easily be duplicated but you'd have to lie about how it was done in order to make the same claims because I can tell you without moving from the couch that you will not be able to replicate this with any camera you have- outside BS'n. And we have no reason to believe that this is just bullshit. Where would RED be if Jim had claimed it had 28 stops of useable dynamic range? Don't you think people would notice this kind of thing when it came time to roll camera?

And this tonal range isn't made up of flare.

[Ace]

That has to be one of the most F**ked up things Ive ever seen and that's saying something. Incredible.

If for any reason its true, RED, Please make it better and cheaper.

[Ace]

By the way, if they are using electronic shutter HDR processing, then this is way easy. Yeah, I know all you geeks have been thinking about the magical sensor that takes 3 frames for every frame and blends it on board. This may just be it.

[J. Eric Camp]

There is another source. It's not even a question.

The filament is clipped so the question is, if the ratio on the bulb around the filament (brightest place I see detail) to the deepest shadows really is 23:1 and if this was achieved by true means. Then comes the need to see a moving image. You can get away with more in a still than in motion.

Is it converted to a 12bit for recording? There are many many questions.

[Anthony Gratl]

There is another source. It's not even a question

It is a question. Perhaps you'd like to explain why it isn't one though, as per thread path.......

[Charles Angus]

23:1 is only 4.5 stops range...

23 stops of range is a contrast ratio of 8388608:1.

Doesn't look like exposure blending to me, that creates solarization in my experience (even if its hard to see). I didn't see any of that.

[Daniel Browning]

With the light pointed into the lens you get lens flare that lifts the shadow areas, in fact if you put bias lights into your RED ONE you could get a A+B=C curve that expands the dynamic range, but would give too much noise in the dark areas since the sensor area is too small.


I'm getting a little offtopic here, but I've seen this claim a lot and never been able to make it work in my own digital photography or understand the principle. To my eyes, the additional veiling flare doesn't add any information (details) to the shadows, it just lowers contrast (as if I had done it in Photoshop). (Maybe I'm doing it wrong.)

I understand how it worked in film: film has a nonlinear response, so flare (or even flashing) caused more of the grains into the higher response range where just a few photons would set them off. Digital, though, is linear, so even at the lowest levels (shadows) the response is the same.

Is it because the flare helps some of the shadow details to get further from the noise floor?

[david farland]

I agree...looks like some HDR process.
If it was 23 stops, looking at it on a 7/8 stop monitor won't show it.
Like Graeme needing to adjust the Red test images in photoshop to reveal 11.3 stops.
anyway, if you import the image into photoshop and look at the luminous curve, you see shadow and a little highlight detail only, but nothing in the middle.
This suggests two images taken a different exposures during sensor sampling or in post.
High & low exposures here:
http://www.reduser.net/forum/uploaded/48_1236038609.jpghttp://www.reduser.net/forum/uploaded/48_1236038648.jpg
Combined exposure here:
http://www.reduser.net/forum/uploaded/48_1236038864.jpg

Still the cleverness of how they did this so it looks a plausible image (except for the mids) is lost on me.
Dave

[Matthew Greene]

On a CMOS sensor you could control the gain of each pixel independently on capture. If this is real, that's what I would guess that they're doing in this sample. The image itself can't prove how many stops of range they're acomplishing without further data. Could be 10, could be 23.

[Dan Hudgins]

I understand how it worked in film: film has a nonlinear response, so flare (or even flashing) caused more of the grains into the higher response range where just a few photons would set them off. Digital, though, is linear, so even at the lowest levels (shadows) the response is the same.

Is it because the flare helps some of the shadow details to get further from the noise floor?

The A2D converter has a threshold. The pixels have a noise level. At low light levels the bias light introduces chaos.

If you modulate the bias light you can imporve the chaos.

In order to extract detail below the LSB the chaos needs to lift the pixel above the lower threshold of the A2D some of the time.

If you can record the chaos on a large enough sensor area and average the pixels you can get values smaller than the LSB of the A2D. If you can average many exposures at the same sensor area you can get values smaller than the LSB of the A2D.

Anything you do to reduce noise will defeat the gains. Did you shoot JPG rather than RAW? Newer cameras do not shoot true RAW data, so you might not record enough of the chaos.

Being linear or not is not the issue, it is the chaos being recorded since photons are integer to all sensors or film.

The larger sensors RED CO. is coming out with could give results of more than 14bits if you reduce the size of the images (with a resize that averages groups of pixels so all pixels are used, not a sub-sample) before compression, and control the chaos in the bias, i.e. white or pink noise in the bias source mixed in with the sensor signal analog before the A2D.

This works the same way dithering an 8bit image at high resolution gives you the tonal range of a 10bit image on the same monitor. If you use compression these tricks will not work because compression removes the low contrast detail.

If you average all the pixels in an image (or a gray card) and make a series of images with the light source brightness increasing very slowly, if all the pixels flip their binary values at the same time you would get stepped brightness changes, but if the binary values of the pixels flip in different frames then the average will climb on an almost continuous slope. If you only average part of the sensor you will see more noise in the slope than you will if you average the whole sensor.

Black and White film records only clear and opaque spots as grain, what looks like the continous tones of gray is the average of many on/off events at noisy thresholds.

[Wolf]

This question came up again so I'm bumping this thread.

[Wolf]

Not sure why so many otherwise smart people seems to think this is a hoax.

A little search around various cinematography forums reveal that several well respected DPs have been shown images from prototype sensors where Thomson Grass Valley used individual exposure control of each CMOS pixel to increase dynamic range. That was around 2007/2008... I don't know if this is the same thing, but that kind of dynamic range is nothing new from these guys.

Might also be a good idea to check the track record of Thomson Grass Valley (formerly Philips, BTS, Grass Valley...) before posting all sorts of conspiracy theories. These are the guys who made the Spirit 4K film scanner, Viper Filmstream camera, invented the Frame Transfer CCD to name a few...

Just saying.