Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

HDR Mode...

I'm quite curious ... about how they implement it.
My idea: Read the sensor output at 1/240 s, but continue with the exposure up to 1/48s, when you read out again.
This way, the highlights would have less motion blur, but that would be the only tradeoff and you'd gain about 5 stops of DR.

I'd like to mention two things:
* It can never be a post trick. You need multiple exposures. With my idea, you have to read the exposure before clipping to gain additional information.
* HDR itself only means more dynamic range. To be able to view the extra-informaition on a monitor or projected you have to reduce the contrast somehow. Because the standard-approach often looks to flat and dull most people increase micro-contrast and saturation to gain this "fake" look. But you could also just let parts of the image clip or use a film-like shoulder to accieve a more natural look.

Regards, Matt
 
Digital Coast Films said:
HDR is quite useful for pulling in all that range. The tonemapping doesn't have to look "Grunge" though. These are all high contrast scenes I "attempted" to make look less so. All took 7 to 9 shots, merged to HDR then tonemapped. My goal to get all that dynamic range into an 8bit image.

I'm extremely excited about Scarlet now :)

dci.jpg
Click to expand...

when epic could do this type of HDR pictures...well i would be speechless.
 
Matthias Hutter said:
I'm quite curious ... about how they implement it.
My idea: Read the sensor output at 1/240 s, but continue with the exposure up to 1/48s, when you read out again.
This way, the highlights would have less motion blur, but that would be the only tradeoff and you'd gain about 5 stops of DR.

I'd like to mention two things:
* It can never be a post trick. You need multiple exposures. With my idea, you have to read the exposure before clipping to gain additional information.
* HDR itself only means more dynamic range. To be able to view the extra-informaition on a monitor or projected you have to reduce the contrast somehow. Because the standard-approach often looks to flat and dull most people increase micro-contrast and saturation to gain this "fake" look. But you could also just let parts of the image clip or use a film-like shoulder to accieve a more natural look.

Regards, Matt
Click to expand...

I thought about that too.

The sensor does a "read/reset" after each frame. It's the M/M-X's fast read/rest times that have allowed for such fast frame rates.

Perhaps part of this is doing multiple reads at shorter intervals before doing a reset, say at 1/250th, 1/125, 1/60th, etc.... all stacking up to a standard 1/48th sec duration. Then reset the sensor and repeat.

While the first and last exposure will have different motion blur characteristics, perhaps it would be ess objectionable than if you actuaqlly reset the sensor between the shorter bracketed exposures.

-sc
 
everyone seems to be fixated on the hardware behind hdr but it's really all about software for me. unless you can further manipulate each exposure "bracket" in post (whether we're talking stills or motion) in a responsive and flexible way, it's just a neat looking gimmick. it's not too different from shooting raw vs. baked-in. not having full control over the image dynamics in post would be unacceptable to most red users today so i would hope red is working on something that let's hdr shooters play with each bracket independently.

in any case, i'm excited to hear about motion hdr.
 
I'm guessing that 3 factors are making HDR possible here... 1st is the faster read rate of the sensor at 5 ms. Second is the 13 stops of dynamic range. So since the sensor can read so much faster it can possibly grab two captures in the same space as it would normally capture one. The third is the RAW workflow.... the heavy hitting is done in post.

In a normal and natural world, the brightest highlight we would need to hold is a white textured wall in broad sunlight ( f16, iso50, 24fps, 48th sec ), and we've seen what RED-X can do with Leo in a match lit scene at iso 2000.... Therefore one could be deep in a cave looking out toward a marble hill side and get deep dynamic range otherwise not possible with a single scan. All done at the sensor and the lens can stay at a happy hyperfocal setting.

It's been a while since I've taken my 16"X20" mahogany camera out to shoot some sheets for platinum/palladium print contacts, but the reality that I could now shoot motion with the same creamy shadow detail where you can see on and on into the shadows under a lofty oak tree gives me a massive broner! Could you imagine a Western film that looked like this?

AAA_ivinwill_9317.jpg

.
10whiteeveningonthecolumbia.jpg
 
HDR mode was my entry for the 'killer feature' contest that Jarred had on here way back when... aww, very fond of those early days. Nice to see it finally in the works. I take full credit Jim.
 
I love that the abbreviation HDR means different things to different people. Some see it as capturing a high dynamic range, some see it as tonemapping. It's amazing the reaction that those three letters provoke. Interesting times my friends....

Graeme
 
Steven Caesare said:
Perhaps part of this is doing multiple reads at shorter intervals before doing a reset, say at 1/250th, 1/125, 1/60th, etc.... all stacking up to a standard 1/48th sec duration. Then reset the sensor and repeat.
Click to expand...
That is exactly what I thought of.
But you are right about one thing: More than 2 exposures wouldn't increase the DR starting from 1/250th, but the transition from more to less motion blur would be a lot smoother.

Regards, Matt
 
Graeme Nattress said:
I love that the abbreviation HDR means different things to different people. Some see it as capturing a high dynamic range, some see it as tonemapping. It's amazing the reaction that those three letters provoke. Interesting times my friends....

Graeme
Click to expand...

And what does HDR mean to you? :D
 
Digital Coast Films said:
HDR is quite useful for pulling in all that range. The tonemapping doesn't have to look "Grunge" though.
Click to expand...

nice, normal, polished HDR like this is what I'm after. Some people say it looks fake? Well, don't use that feature....

I personally can think of many uses for it especially composited with some 3D.
 
Roberto,

For sure HDR is a cool tool to have at your disposal, and by no means am I against that look, and dabble with it myself. I think it is interesting and unique, but should not be the only way to apply HDR images to your work. I just don't think RED is going to provide some kind of in-camera tone mapping tool (maybe they will include it in Redcine-X?).

I also have a hard time wrapping my brain around what an excessively tone mapped image in motion would look like :blink:
 
HDR images with just 2 exposures can help you bring back the clipping you would have from a single exposure. For example table lamps blowing out or windows behind subjects.

It doesn't have to look fake, shooting normal exposure and another 2 stops under can be enough to recover highlight detail.

That's what excites me the most here, getting more DR out of the sensor for high contrast situations.

Very interested to hear how it works for motion, would that not create ghosting on shots with high movement due to the 2 or 3 exposure delays?
 
My guess is they are shooting 1 frame in Raw, then Processing it out to 2 or more exposures and then combining them for output. Just requires fast cpu and lots of math.
 
Hi Ben,

Shooting 1 exposure in raw is what is currently done anyway. Redcode is raw. It would limit you to 12/13 stops of DR of the MX sensor, no matter how you process that frame. Thats not HDRI. If highlights are blown no processing will bring those back. If you raise the shadows you will see noise just like pushing a curve in redcine.

The camera would have to shoot at least 2 different exposures to increase the DR for HDRI
 
1 year ago I posted an image of a bayer pattern with HDR capabilities. Wonder if it will come alive in red.

Shooting extra frames for different exposures just wont cut it, for anything beyound dead landscapes. You need to sacrifice some pixels to make it work. Why do we need twice as many green?
 
Nik Skjøth said:
1 year ago I posted an image of a bayer pattern with HDR capabilities. Wonder if it will come alive in red.

Shooting extra frames for different exposures just wont cut it, for anything beyound dead landscapes. You need to sacrifice some pixels to make it work. Why do we need twice as many green?
Click to expand...

yes that what I thought, would only be useful for static shots.

Since Red go crazy over line skipping I am really keen to know how this will work for motion. Could you even bracket with alternating lines? A form of HDR interlacing if you like. Tho it would lower resolution....

Im done guessing. They'll tell us eventually!
 
Different lines for different exposures would be the straight forward thing to do, but as you said. Jim would introduce lineskipping over his dead body.
 
Graeme Nattress said:
I love that the abbreviation HDR means different things to different people. Some see it as capturing a high dynamic range, some see it as tonemapping. It's amazing the reaction that those three letters provoke. Interesting times my friends....

Graeme
Click to expand...

Well I know that you know which is the real one. So I'm not worried. :D
 
Last edited:
You must log in or register to reply here.
Back
Top