4k HDR Cheat for Scarlet- Is it Possible via ISO Metadata?

[Ryan Sauve]

I'm sorry if this has already been discussed and/or it's a dumb question but couldn't you cheat 4k hdr via the use of 2 tracks of the same shot using different ISO levels since ISO is metadata that isn't burned in? Has anyone tried this and if so, any best practices to share?

 

[Akos Nyoszoli]

If you really lost detail either in the shadows or in the highlights, you can't bring them back by changing the iso. It doesn't work like dslr iso, it works more like analogue film: if you didnt expose anything on the negative (say, you shot many stops under), you cant get detail there just by increasing the development time (the iso), because there is nothing to amplify.

 

[Akos Nyoszoli]

So the sensor has one native iso (to me it seems like 320), and you cant change film stocks. You can push it a bit (to 500 or 800), but its really just pushing, not changing stocks.

 

[Steve Phillipps]

I'm sorry if this has already been discussed and/or it's a dumb question

No such thing as a dumb question, but no. In fact to do what you're suggesting is just standard grading, you don't need 2 tracks - you pull up detail from the shadows and bring down the highlights separately all in the same shot.
Steve

 

[Mike Lary]

I'm sorry if this has already been discussed and/or it's a dumb question but couldn't you cheat 4k hdr via the use of 2 tracks of the same shot using different ISO levels since ISO is metadata that isn't burned in? Has anyone tried this and if so, any best practices to share?

If you buy a half pint of milk and pour it into a pint glass, you still have a half pint of milk. Changing your metadata will only manipulate the data that's already there. It can't create data above the range that was captured. You are bound by the sensor's limitations. HDR allows for compositing of bracketed exposures in order to extend the dynamic range of your image with actual recorded data.

 

[Christopher Barrett]

A lot of digital Raw formats don't process out all of the information they capture. The ISO sort of restricts the available range. I do this all the time with stills when I need more DR in a shot when I can't bracket, like with people in it.

This is my approach for stills... Find a good middle exposure that doesn't clip as your Raw capture. Process it once at + 1.5 stops and once at - 1.5 stops (that's as far as I can go with my Phase one files). Combine the two with a luminance mask and voila, extended dynamic range.

Now, from my work, I believe that you CAN actually get all of the information that a RED file contains out of the RAW and into a single file without having to do all of the above. Usually I just drop down to RedLogFilm, dial in some contrast and there ya go.

Damn. I wish I had Stills software that would extend the DR that easily.


GO test it and let us know

CB

 

[Nick Pasquariello]

The way HDR works, it takes two DIFFERENT exposure lengths, with the X track being an exposure that takes place sometimes in the middle of the A track.

Think of it this way: let's say you have an exposure time length 1 second (you wouldn't, but for the sake of understanding HDRX, let's say that we do.) For each frame, the A track is exposed over the course of 1 full second. Frame starts, and for the entire next second, the light that hits the sensor is recorded to the frame of the A track.

The corresponding frame of the X track, however, doesn't start yet. Let's say that the X-track exposure is .5 seconds. Then when the frame of the A track starts recording, wait .25 seconds, and then the frame of the X track starts. .5 seconds then pass (durning which data is being gathered for BOTH the frame of the A-track and the frame of the X-track). At this point the frame of the X track is finished, a full .25 seconds before the frame of the A track.

See, for HDR to work, you have to have different amounts of source light. Whether that's aperture or shutter speed. If something is clipped in your A track, you have a good shot of still having that data in the X track. In any instance where you truly NEED HDRX, just changing your ISO and exporting out a different file will not actually help you, as that data was blown out and gone forever.

 

[Eric Haase]

A lot of digital Raw formats don't process out all of the information they capture. The ISO sort of restricts the available range. I do this all the time with stills when I need more DR in a shot when I can't bracket, like with people in it.

This is my approach for stills... Find a good middle exposure that doesn't clip as your Raw capture. Process it once at + 1.5 stops and once at - 1.5 stops (that's as far as I can go with my Phase one files). Combine the two with a luminance mask and voila, extended dynamic range.

Now, from my work, I believe that you CAN actually get all of the information that a RED file contains out of the RAW and into a single file without having to do all of the above. Usually I just drop down to RedLogFilm, dial in some contrast and there ya go.

Damn. I wish I had Stills software that would extend the DR that easily.


GO test it and let us know

CB

Absolutely spot on here. While the majority of people will tell you no, you actually can squeeze a bit more out of the image by "developing it" at 2 different iso's and keying highlights. I've done this on some commercials I shot on film where we dialed in a look we like in telecine that is fairly contrasty. The problem is that this look would blow out the sky, even though detail was retained on the negative. The people looked good, and just bringing the highghts down muted the overall image. So we often lay down a separate sky pass and key it back in. It's pretty common. With red, you can encounter a similar situation where a grade or ISP development will put the image in a place that is good for SOME tones of the final image but not all. You can develop a second track at a lower ISO and regain some of the detail in highlights that appeared blown out on the initial iso development. Sometimes, you can just bring down the highlights on the initial iso depending on the grading app to get this info back but sometimes there is a lot of overlap to midtones and you might not like the final image. So yes, sometimes you can squeeze a bit more out the raw data with this technique.

 

[Ryan Sauve]

Wow guys, thank you so much for all this info. It makes sense and I feel like I learned a lot today!

 

[Jaakko Rinne]

Absolutely spot on here. While the majority of people will tell you no, you actually can squeeze a bit more out of the image by "developing it" at 2 different iso's and keying highlights. I've done this on some commercials I shot on film where we dialed in a look we like in telecine that is fairly contrasty. The problem is that this look would blow out the sky, even though detail was retained on the negative. The people looked good, and just bringing the highghts down muted the overall image. So we often lay down a separate sky pass and key it back in. It's pretty common. With red, you can encounter a similar situation where a grade or ISP development will put the image in a place that is good for SOME tones of the final image but not all. You can develop a second track at a lower ISO and regain some of the detail in highlights that appeared blown out on the initial iso development. Sometimes, you can just bring down the highlights on the initial iso depending on the grading app to get this info back but sometimes there is a lot of overlap to midtones and you might not like the final image. So yes, sometimes you can squeeze a bit more out the raw data with this technique.

But why on earth would you make a second ISO pass for this, when you can just do a secondary for the sky? If you're using RedLogFilm to bring your RAW footage into grading, you shouldn't be actually clipping anything yet. So just do your high contrast grade and key the sky to bring it back down with a secondary.

 

[Eric Haase]

But why on earth would you make a second ISO pass for this, when you can just do a secondary for the sky? If you're using RedLogFilm to bring your RAW footage into grading, you shouldn't be actually clipping anything yet. So just do your high contrast grade and key the sky to bring it back down with a secondary.

Depends on the grading app. Most grading apps behave this way. But if you were to try this in Apple's Color, which is EOL but still used a bit the secondary wouldn't get the info back. So in theory yes you could do it, but in practice you probably won't need to most of the time.

 

[paulherrin]

it's really a preference/workflow thing... you can go redlogfilm, or otherwise contain your raw histogram and grade to your liking - or you can combine multiple passes in any which way. less post processing math = technically better, but it takes more work to combine passes correctly (for motion). honestly, it would be very hard to notice any quality differences. so really, just do whatever you're most comfortable with depending on the shot/project/software requirements.

one question is whether the ISO change does more than simply shift the histogram... either way, it should be technically better to, say, lift shadows via ISO/FLUT/in-the-raw than it would be to apply a levels/curve adjustment. however, you probably wouldn't notice a technical issue unless you're working in 8-bit - which i don't recommend.

if you don't really know what you're doing, and have the luxury - get it as close as you can to the final look in redcine/raw. if you do know what you're doing... well, good for you.

 

[Jaakko Rinne]

Depends on the grading app. Most grading apps behave this way. But if you were to try this in Apple's Color, which is EOL but still used a bit the secondary wouldn't get the info back. So in theory yes you could do it, but in practice you probably won't need to most of the time.

In Color you'd probably want to first bring the sky down in a secondary and then do the high contrast with primary out or another secondary (or depending on the shot, bring the sky down inside the matte in the first secondary and do the high contrast outside the matte... or vice versa). So you wouldn't really need two passes even there.

And on the other hand, is it even possible to use two passes in Color? You'd need to composite them outside of Color, right?

 

[Jaakko Rinne]

one question is whether the ISO change does more than simply shift the histogram... either way, it should be technically better to, say, lift shadows via ISO/FLUT/in-the-raw than it would be to apply a levels/curve adjustment. however, you probably wouldn't notice a technical issue unless you're working in 8-bit - which i don't recommend.

My understanding is that ISO adjustments with RED footage compress the high and/or low end to avoid clipping. The further to the left or right you push your histogram with ISO/FLUT, the more the values at the extreme you push it to get compressed. With other adjustments it's easier to clip values.

What I tend to do is get WB/ISO/FLUT to be where I want it and leave the curves etc. for the grading app.