"HDR MODE" - Possible on Red?

[Joe G.]

We talked about this before, but I never heard a response from anyone at the company.

As most here would know, HDR combines 2 frames to make one frame with expanded dynamic range (more highs, more lows) by bracketing the exposure of consecutive images. It occurs to me that this is feasible, since most people shoot 24 fps, and use a 180 degree shutter. If you record at 48fps, there is the possibility of using the two frames and putting them together into one intended for 24 fps playback.

Here's my suggestion:

HDR for 24P out of 2 combined frames:

2 F stops over bright protection

48fps at 360 degrees (1/48s)
48fps at 90 degrees (1/192s)

or

1 stop over bright protection

48fps at 360 degrees (1/48s)
48fps at 180 degrees (1/96s)

Some piece of software can combine these quickly into 24P, with varying mix settings.

Anyone there done tests?

Other manufacturers will no-doubt step up and give this a try if Red doesn't.

There is also the FUJI idea of using photosites of various sizes across the array. The larger sites gather more light than the smaller ones. Anyone at Red considering this approach to expand dynamic range?

Now, this doesn't have to work great in every shooting situation, as in walking along with dialogue. But, the option would be great for landscapes, or scenic shots.

 

[Noah Kadner]

Hmm- I've seen this proposal at least a dozen times on the forum- no offense. I think the problem comes down to temporal differences frame to frame that would prevent reliable motion rendering. Not to mention it would require incredibly fast and accurate control of the mechanical iris to really be effective and that would require whole new lenses to be designed and built. Your shutter method wouldn't do it because the look of the two different shutter modes would make combining the frames all but impossible on fast motion. HDR motion picture cameras will come someday- probably sooner than we think but I don't see it on the RED ONE.

Noah

 

[Kim Frank]

Why don't you just compose two different rendered ISO settings like in RAW still Photography?

 

[David Mullen ASC]

Combining two sequential frames into a single frame would create motion artifacts since a moving object occupies different positions on each frame. Plus since the whole idea is that one of the two frames is darker than the other to gave more dynamic range, that moving object would alternate in density for the two positions it exists in. Imagine a shot of two people tossing a white baseball between them...

It's similar to early color films where they tried to take a green and then red frame sequentially -- even at higher frame rates, there was color strobing.

--

Using two different ISO settings in post doesn't gain you more information than what was captured in RAW.

 

[Pawel Achtel]

It can work with two sensors and a part-mirror.

 

[Dan Hudgins]

Pellicle Beamsplitter

Pellicle Beamsplitter

Use a Pellicle Beamsplitter and two cameras in sync, you only major limit is ho much over exposure the dark camera can take before the high end Solorization becomes an issue.

I thought I saw some "black" headlights in CHE#1, but other than that I think you should be able to get some stops of extra range.

You can align the images by sub-pixle then reduce them back to 4K in software, not something hard to do.

You can read my many older posts on this subject.

==

Right now you should note that there is "DRX" in the de-mosaic software for the REDONE that will give you some extra highlight detail that is off color maybe, it uses the clipped part of the highlights to put some detail back in even though it cannot know what color that luma datail is for sure, or something like that, just look for the "DRX" option.

 

[Mohammed El Sharqawy]

maybe faster rates would do the job say instead of 48fps use 96fps .. skip frames, simply this would give you half of the artifacts that is present using 48fps.. and as Joe G. mentioned its not for all shooting conditions.. lanndscapes maybe timelapses are perfect applications.. coz there isn't much motion with such shots.. plus the given artifact will be something like motion blur which you will need to add if shooting high frame rate/shutter speed.. compensating with shutter speed is the best way to go around coz apertures are mechanical and can't work out that speedy open/close...
also the idea of using different photosites is great.. but this needs the ability control shutter speed for each pixel to arrange difference in exposure... which i think would not be possible when dealing with normal cmos and ccd coz they read sequentially don't know if its possible to map pixels..FUJI got some techs for doing this like Double exposure (NEW) different pixel sizes (OLD)..
and with epics doing so would be better and better advantage.. you got some nice speed ups in frame rates which artifacts would be lower and lower.. say you can reach 240 fps then skip 2 frames each 5.. less time gap,, less motion trouble..
I had an experience before usshooting HDR timelapse ing Canon EOS 450D, Shooting RAW, 3 Exposures/frame.. connected to a laptop recording each exposure in say 5 secs.. that gives 5 secs difference between 2 exposures.. and the results was nice.. I may post it soon..

 

[Vico Martin]

Using two different ISO settings in post doesn't gain you more information than what was captured in RAW.

I agree not with shutter not with ISO but...

with two gammas?

I mean keeping i.e. the detail level of a dark gamma for overs and a logaritmic for subexposures and apply both....

IMHO this is the unexplored depths of RAW, and some day the colorist apps allows to apply a diferent gamma in a secundarys room or something to recover the unused data.

Will see. This is the software age.... and Graeme is everywhere. :emote_couch:

 

[Michael Hastings]

It can work with two sensors and a part-mirror.

And it seems like it could double as a stereo 3d camera as well -i.e. HDR or 3D

 

[David Mullen ASC]

I agree not with shutter not with ISO but...

with two gammas?

I mean keeping i.e. the detail level of a dark gamma for overs and a logaritmic for subexposures and apply both....

IMHO this is the unexplored depths of RAW, and some day the colorist apps allows to apply a diferent gamma in a secundarys room or something to recover the unused data.

Will see. This is the software age.... and Graeme is everywhere. :emote_couch:

Why are you assuming that RAW contains more information than you can currently access in post color-correction? RAW just means all the data off the sensors was captured, it doesn't mean that the RAW image is naturally HDR. Believe me, there is no extra information there if you are doing the color-correction process correctly. You can use all the gamma controls you want but you are still working with a set range of values. It's not like there is a hidden 16-stops of dynamic range waiting to be unburied.

RAW isn't magic.

 

[Radoslav Karapetkov]

Use a Pellicle Beamsplitter and two cameras in sync, you only major limit is ho much over exposure the dark camera can take before the high end Solorization becomes an issue.

I thought I saw some "black" headlights in CHE#1, but other than that I think you should be able to get some stops of extra range.

You can align the images by sub-pixle then reduce them back to 4K in software, not something hard to do.

You can read my many older posts on this subject.

==

Right now you should note that there is "DRX" in the de-mosaic software for the REDONE that will give you some extra highlight detail that is off color maybe, it uses the clipped part of the highlights to put some detail back in even though it cannot know what color that luma datail is for sure, or something like that, just look for the "DRX" option.

This is an interesting concept, I read a similar thread on Scarletuser.

I wonder if such a two-camera HDR rig can come close to the latitude of film?

 

[Joe G.]

Well people ARE interested in some solution to get HDR. I wonder if Red is actively working on it?

 

[Jason Murphy]

I would personally be surprised if they weren't working on it in some capacity (as in a few people asking "Can we make this happen in a new camera down the line, and is it worth it?"), or doing some research into the best way to go about it.

I would be very surprised, though, if a proper HDR solution were ever going to be available for this round of Scarlet/Epic. And by proper, I mean, one that doesn't involve motion artifacting or an entire redesign of the camera.

 

[David Mullen ASC]

True HDR is beyond the capabilities of a single sensor cine camera -- at best, what you can do is keep trying to improve dynamic range while reducing noise (making more of that range usable). But to achieve what true HDR photography allows, 20-stops or more of dynamic range, is not really possible for a single exposure at a single point in time on a single sensor, unless you get into some trick of putting more photosites on a sensor some of which are dedicated to overexposure detail. But even that would at best add only a couple more stops of dynamic range.

The question is how many people actually need HDR information to take a picture? How many of you DSLR shooters combine multiple exposures of the same image to create HDR photos? How many cinematographers want 20+ stops of dynamic range in the picture? HDR seems to be a niche sector of photography. It would be more useful if a digital cine camera could just match color negative's dynamic range, which is more in the 14 or 15-stop range at best.

 

[jimhare]

What about some kind of mad scientist spinning ND in sync with the shutter?

 

[Andrew Wilding]

Frankly, I wouldnt even want more then 15-16 stops of DR. I don't want photographs to look like the real world, to exactly mimic what the eye sees - photography is compelling because it is a more limited slice of light/life than we normally get - the art of exposure is choosing which slice of light to use. I think alot of the magic, fantasy elements of storytelling/moviemaking come from the limitations inherent to the medium. Im not rushing to see them go away.
I think its very easy to fall into a trap, thinking that new technology will somehow turn you into an artist. Lets not put the cart in front of the horse. The technology serves the artist. The brush doesn't make the painter.

That said, I am myself excited by all the huge technological leaps taking place. Digital is getting better everyday, and if not for that fact, I dont think I would even be entertaining the notion of being a film maker. Thanks, Red,

 

[Mohammed El Sharqawy]

never seen 15 stops on any medium before

never seen 15 stops on any medium before

The question is how many people actually need HDR information to take a picture? How many of you DSLR shooters combine multiple exposures of the same image to create HDR photos? How many cinematographers want 20+ stops of dynamic range in the picture? HDR seems to be a niche sector of photography. It would be more useful if a digital cine camera could just match color negative's dynamic range, which is more in the 14 or 15-stop range at best.

I've shoot lots and lots of HDR.. and by time it became a habit for me to shoot most of my images in three compensated catpures +/-2 stops.. later in post I got all the dynamic range I want then I produce what image I like.. low or High DR.. having more DR is always advantage and saves lots of @$$'$..

I've never seen a DP using Vision2 500D as the main stock for a movie.. or I mean shooting the whole movie on that stock .. which is about 12 f-stops of latitude.. ASA 500 is not the all in all film stock to use in all situations.. it adds some nice grains to the image.. vision3 rumored 14 f-stops not 15.. but actually I never tried it.. reaching such dynamic range with kodak best stocks needs extreme conditions about lighting in front of the lens and develeping it after.. say your lab changes the chemicals on each reel you develop.. not even every 100 reels.. most 100D and 200D stocks (they give a nice fine grain image not like the 500D) ranges from 6 to 9 f-stops... film is out of this compition.. FUJI DSLR sensors (I mean SuperCCD SR) got dynamic range in RAW files which I've never seen in any medium before.. neither RED nor any other DSLR nor FILM.. (I noted I never tried vision3).. they just implemented a technique which increases the dynamic range by changing photosites size.. later they added temporal value to do the job..

 

[Dan Hudgins]

Pull-process

Pull-process

I've shoot lots and lots of HDR..

You have to pull-process the footage to avoid too much crush in the highlights.

You can also do something like what was once called "chem-tone" I think, you process the color negative in a special black and white developer that uses a special developing agent that limits the highlight density, then fix and bleach then develop in color developer then fix and bleach again, the end result is a very flat negative. I think "chem-tone" was used for a two stop push to avoid blocked up highlights, but if you leave the Hydroquinone out of the first developer and just use Phenidone A you get less speed but a longer scale.

Older negative stocks had more silver and so were more soft on the sholder, newer thin films have less silver and can be more "brittle", even so pull-process 2 stops or more can help.

You can also process color and black and white negative films reversal using fine grain negative developers as the first developer to get softer results, something like "chem-tone" but you get a positive rather than a negative to scan.

You can also re-develop twice to get even softer results, but probably too soft, for the most part a 2 stop pull-process will extend the highlights enough, you over-expose when you pull process so that you do not loose the shadow detail, but how much you over-expose takes some tests.

I have rated 800 speed film at 1.5 and gotten usable results with finer grain.

BTW: Older movies negatives look higher contrast in new prints because todays printstocks do not match the contrast of the older negatives, todays negative stocks are lower contrast to reduce grain at high speed and todays print stocks are higher in contrast to make up for the lower contrast negatives. The 70mm re-print of 2001 was way much to high contrast and did not match the original film at all, also other "restored" moves are not at all the way the older print-stocks looked at the time of first release. If you can try to see "vault" prints that were held and not projected so you can see better what the original prints looked like.

 

[Dan Hudgins]

Inverted highlights issue...

Inverted highlights issue...

I wonder if such a two-camera HDR rig can come close to the latitude of film?

You can shoot some tests with a DSLR and see. My software can fuse images.

The big issue is inverted highlights, I noticed on this web site that what seems to be inverted highlights on the CMOS sensor camera and not in the CCD sensor camera?

You can see, look at the lamp over the sink, does the light look Solarized so that you can see what looks like a CF lamp in the CMOS camera but in the CCD camera the lamp is burned out?

http://www.davidrubio3d.com/view_topic.php?id=481&forum_id=116

See the image file named, 720p6mm1.2FRAME_8-Gamma_curves.bmp, could be something he did in CC, but looks more like the highlights are inverted in the lamp so the bulb is darker than the reflector.

The only time I think I may have seen that with the RED ONE was on the car headlights in CHE#1 I think, so if RED ONE does Solarize that would put an upper limit on the exposure, and the REDCODE noise level would set the lower limit, between those might be 14 to 16 stops maybe more, someone would need to do some tests.

 

[Chris Clogg]

It'll be interesting to see where cameras are going, cause I think eventually we won't need more then a certain amount of pixels... which is where we branch into HDR and maybe capturing several focus distances (plenoptic camera or w/e it's called)... so in post-prod it gives you the ability to refocus completely, and re-exposure completely... Lot of data.... Really though, we can't say what will be here in 20 years... it'll probably be better than anything we can imagine.