Knowing and 2k

[Dan Hudgins]

DRX in REDCINE

DRX in REDCINE

Dan, thanks for the reply. Could you elaborate on 'highlight extension'?
Nelson

There seems to be a DRX feature in REDCINE,

http://www.reduser.net/forum/showthread.php?t=21710

The sensor in REDONE seems to be exposed to produce unbalanced images in Daylight, meaning the probably the green sensors clip before the red sensors, and the blue clips least (or manybe green, then blue, then red).

At any rate, you do not get 12bits dynamic range since to get white to clip in all three colors you need to clip the other two down to the bit depth that matches the white brightness, so you get like 12bits green, 11bits red, and 10 bits blue.

Because the blue or red will take more light to clip, you can extract the image data above the equal clip point and make it monochrome then combine that with and above the color image to get some detail back in the highlights, kind of like HDR images.

You would get lower noise if you shot with a pink filter on the camera and adjusted the de-Bayer to get 12bits in R, G, and B, but you would not be able to extend the highlights in monochrome since all three colors would clip at the same brightness.

I was not so much concerned with what was outside the windows being blown out as the effect of the high brightness on the actors faces, in that case gelling the windows and using more fill would reduce the lighting ratio and let the exposure go down to keep color balance on both sides of the actors face better perhaps...

 

[Rich Schaefer]

Looked a bit soft in the wides.

Perhaps it was a lens thing, wide lenses are way more critical on flange-focal adjustment ie: backfocus than longer lenses. (It is harder for lens techs to setup/back-focus wide than long lenses! Yes, for AC's it's the opposite: to shoot with wide is easier....

 

[Obin Olson]

soft wides on film only looks awesome in digital

 

[filip kovcin]

i am going to see this movie tonight, but i am 100% sure, that some unsharpness in 35mm projection is because of transfering to 2K not to 4K. i tested that with various films comparing digital and "normal" film projections, and ALWAYS wide shots were kind of unsharp, without details etc.

just ONCE i did NOT notice that artifact - when saw black & white film copy transfered to 2K.

just a thought.

filip

 

[Nelson Goforth]

... you do not get 12bits dynamic range since to get white to clip in all three colors you need to clip the other two down to the bit depth that matches the white brightness, so you get like 12bits green, 11bits red, and 10 bits blue...

I have not played with the DRX feature, so thank you for the explanation. That might explain the milky quality I was seeing in sunlit skin: lacking in detail. But wouldn't the in-camera histogram or the 'stoplights' show clipping on one or two channels in that case?

From what you've said there, assuming we want to keep the hot highlight and contrast (i.e. not add more fill), one solution is to lower exposure a stop and adjust in timing. That a critical detail in a scene (the actor's face) should not be at the extreme right of the histogram?

I could see that working in a bright daylight scene, but in a darker scene with one hot key hitting an actor - and in which we want to see into the gloom around them - I'd lose the detail in the shadows.

Again, I'm going on one viewing at one theater. I'll have to set up a few similar shots and try it with our camera to see what happens.

Someone in this thread said that Knowing was a 2K transfer to film. Do we know that to be the case, and would that have a bearing on this topic?

Thank you,
Nelson

 

[Dan Hudgins]

DRX vs. fill light

DRX vs. fill light

From what you've said there, assuming we want to keep the hot highlight and contrast (i.e. not add more fill), one solution is to lower exposure a stop and adjust in timing. That a critical detail in a scene (the actor's face) should not be at the extreme right of the histogram?

I would think the green histogram would be at the high end, but something is always off the high end of the histogram in shots with sunlight since any Specular reflection will have about the brightness of the disk of the sun. I noticed some talk in other posts that the histograms do not seem to display true raw readings in all "color space" settings, rather that show a histogram of what is on the monitor viewfinder, and so if that is true would not display the image data in the DRX zone?

That's why I thought they should gel the windows and use more fill light. Outdoors use more fill light, ND and maybe stop down a bit. In the old days they used ARC lights outdoors for fill, or at least "9 lights".

Someone in this thread said that Knowing was a 2K transfer to film. Do we know that to be the case, and would that have a bearing on this topic?

It could, you want the highlights in the straght line portion of the printing negative, but if you are running the film recorder too fast, or the CRT/light source is old, you can under-expose the printing negative or master positive. Under/Over exposure of the printing negative or master psoitive may give some blocked up off color highlights or tracking errors etc.

Also LCD may have a hardtime with the ends of the tonal range, so if the film recorder uses a LCD that might be part of the issue. In my tests with my film recorder program I found I can get better highlights and shadows off a LCD monitor using three passes (9 total for color) where you adjust the gamma, contrast and brightness for each pass to put part of the image into the "sweet spot" of the LCD monitor, the film then mixes the three passes to give a smoother tonal range. I'm still testing the multi-pass film recorder method, and making adjustments to optimize the image quality. When doing your own DI you can take some extra time to get better results maybe...

Softness can come if the lens in the film recorder is opened up to speed up the exposure, or stopped down past f/8 because the stock used is too fast. The lens could also have been out of focus in the film recorder, or if they made printing negatives from the master positive made in the film recorder on an optical printer that could have been out of focus or its lens not at prime stop to get the work done faster.(?)

 

[Chris Swinbanks]

Don't think too deeply over some of this stuff... remember this was nearly all Build 15, apart from the Boston shoot on Bld 16.
4k (mainly) Red, 2k FX (Animal Logic and a couple of other companies) & 2k DI (Park Road Post) recorded out through an ArriLaser (Weta Digital). Don't know the actual film-out details (stock, formatting, etc) , or even if all negs were done at Weta, or if it went I/P - dupe Neg (although I presume it would have), etc.

Chris


just noticed Sanjun posted in another thread, from AC - Simon Duggan's article..
http://www.reduser.net/forum/showpost.php?p=394333&postcount=258

 

[Dan Hudgins]

Thanks for the info...

Thanks for the info...

2k DI (Park Road Post) recorded out through an ArriLaser (Weta Digital).

Thanks for the info, it helps to know what one is looking at and how it was done, there are so many options now, and each has its own effect on the final look of the film.

or if it went I/P - dupe Neg (although I presume it would have), etc.

Given the number of prints I would also think they made a I/P in the film recorder then several printing negatives. If the I/P is under-exposed you get reduced contrast off color highlights in the print. Master positives made in film to film printing were made quite dark, traditionally, to avoid the crushed highlights problem.

 

[Chris Swinbanks]

I'll try and find out if they did an IP on the Arrilaser (I'm not aware they did...). I never liked to because IIRC, unless setup properly with some overlap during the record, it gives you clear framelines on the IP, black framelines on the dupe neg, and clear framelines on the print.... not so nice, particularly for a scope finish.

 

[Dan Hudgins]

Clear framelines and IP or no-IP

Clear framelines and IP or no-IP

it gives you clear framelines on the IP, black framelines on the dupe neg, and clear framelines on the print.... not so nice, particularly for a scope finish.

I have thought about the clear frameline issue for shooting 16:9 off a 4:3 LCD monitor with my DANCINEL.EXE (tm) program. Since I am going direct to print stock much of the screen would need to be full white to get black frame lines on the print, that tends to flare the lens, and mush up the image area. With Scope you have less area that you need full white to make the smaller black frame line.

I am not so sure clear frame lines need to be a No-No any longer since change-overs are not done, so long as the show starts off in frame it should stay that way using platters. Also polyestar does not break much, in the old days of cellulose triacetate base you would get prints that had 3 perf splices so the reel would go out of frame part way through until the projectionist turned the framing knob. I have not seen the projector go out of frame at a multi-plex recently.

Clear frame lines is also "green" since you use less chemistry up for the 1.85:1 prints.

==

It will be interesting to hear if they made a IP then dupe negatives off that, or just made several printing negatives each in the ARRI laser film recorder.

I saw the roadshow print of CHE and SS said that it was done on a Cinevator, when the projector was in focus it seemed a little sharper in places, which you would expect since it was 1st generation, rather than 2nd or 3rd in Knowing(?).

 

[Chris Swinbanks]

Confirmed, no digital IP's, only digital neg.
The other issue I was reminded of with a digital IP is that they are tricky to QC without running a dupe neg and print (huge cost for a full feature), although I have run IP's through a Steenbeck before, as do some labs for QC'ing.

... and I don't know I'd be game to rely on all projectors having accurate 1:85 gate masks to hide the white light surging around "clear" framing on a 1:85 print... but thats just me... ;-)