16mm 200T and 250D workflow

[Elias Nemer]

Hello,

I am currently working on my first film shot on film. with a Bolex film camera with 16mm 200t for interiors and 250d for exteriors.

The deliverables i will be receiving are
4K Log footage
Best Light footage

I wanted to ask about the process. I've already had a look at the film from a youtube video, I took it and graded it just for fun to see what i might like. obviously just to get ideas. As i was working on this footage it did at least look like log footage as i know it from digital. maybe more contrasty but not enough for me to consider it properly processed with a healthy amount of contrast balance and saturation. Except on the interior scenes which were allot lower in luminannce, very dark, so i guess naturally its might be somehow better exposed because of the darkness as the exteriors seem overexposed so bringing printer lights down a few points helps it get to a good place. (just to describe the film as perhaps it might help determine what and where it is in the process from experience). perhaps it was developed with printerlights set for the interior scene therefore making exterior overexposed. is this possible?

So anyways this film was scanned and developed for free, i wanted to ask about the process. Is there no print? so the log footage i am receiving is a the positive without a print?

In terms of the look i found the best look was just a simple contrast curve to retain the qualities of the film, then just some accurate grading underneath.

also what is best light footage?

Best,

Elias Nemer

 

[Perry Paolantonio]

A print requires a photochemical process (timing, contact printing the negative to print stock, developing that print). It's something some labs still do but not like it used to be where every city had at least a couple film labs and it was relatively cheap to do. Now, you scan the film and work with it digitally most of the time.

Best light is a scan where they do some grading during the transfer to get a basic grade. On modern film scanners you shouldn't do this because it bakes that color correction into the scan and you may not have as much flexibility later, as you would with a log scan of your neg.

Our usual starting point for color neg is basic lift/gamma/gain adjustments in Resolve going by the scopes. We never use LUTs for this kind of thing. Once you've done it a bit you can get a balanced image very quickly. Really helps to have a control surface for this, even a basic one. working with the UI and a mouse is tedious. No matter what, it's a shot by shot thing, but often (in Resolve), once you have a scene in the ballpark, CMD= and CMD- get you most of the way there and you can bang through a reel pretty quickly.

 

[Elias Nemer]

Thanks man, this makes allot of sense. Yeah ive got my mini panel. Literally all the film needs is some offset adjustment. some sat and a curve. then power windows etc. but no luts used, and not doing anything special to the film as it already is what it is. has has the special quality, so i think im on the right track as i have a decade of experience grading just not film. just a question though (im not going to use this for this project as i dont think it needs it) but if i were to use for example filmbox and use the print section of the Plugin, would this be relatively a good idea? or would you advise against it?

also while in here if i could ask you what the grey point of film would be though your experience, ive read maybe around 0.45, also i wanted to ask what is the best color space to consider the film. Cineon log?

 

[Perry Paolantonio]

I've never use that plugin so I can't say, but my experience has been that that kind of thing never really works properly. Every scan is different, even scans done on the same hardware by different labs are a little different because of the settings they may use as standard. It's not much different than back in the day when you'd have prints timed at a lab - the timing lights from one lab are basically meaningless to the next, because they're different setups with different hardware and different procedures. I love it when someone tells me to use the paper punch tape from their film that was timed in 1975 at a lab that no longer exists, as the basis for what they're looking for... Um, yeah.

We primarily work in Rec709 because there's really not much advantage beyond that. Makes it easier too - a lot less to have to fuss with.

 

[Elias Nemer]

Lol yeah, I mean with digital some less color science knowledgeable cinematographers will give you a LUT from an old project and expect to see what they have in their heads. just doesn't work that way.

Thanks for the advice man, i appreciate it allot. makes allot of sense. Thank you!

 

[Dean Winkler]

At the risk of offending the professional colorists on this forum—of which I am not one—here is an alternative workflow you may want to consider:

1. Use the lab’s best-light transfers strictly as dailies to evaluate what you are capturing on set. Stay focused on production and avoid getting distracted with post-production at this stage. Ensure you have three copies of all lab-delivered files, with at least one copy stored in a physically separate location.
2. After principal photography wraps, re-transfer and color correct all of the original 4K log footage to 4K, ProRes 422 HQ, Rec.709 files. These passes do not need to be perfect or absolutely consistent shot-to-shot; they simply need to be high-quality images with no clipped highlights or crushed blacks, and generally within the aesthetic range you are targeting. If script notes were taken against timecode from the best-light dailies, confirm that the new transfers preserve matching timecode. Take your time with this step—do not rush this. Enjoy your footage.
3. Edit directly with the 4K ProRes HQ files —no proxies required. Output the cut as a 4K ProRes 4444 Rec.709 file with a temporary or rough audio mix.
4. Once the cut is fully locked, perform a shot-by-shot final color correction on the edited ProRes master. This is where you can be highly precise about individual shot treatment and overall consistency. For this stage, you may wish to hire a freelance colorist or if you can, work through a post facility.
5. Complete the sound mix and conform it back to the color-corrected picture master. Use this final master to generate all required distribution deliverables.

I recognize this is a somewhat old-school approach, but if you are willing to invest the time, it can produce very strong and reliable results.

 

[Dermot Shane]

in my workflow typicaly delivering HDR home as the hero master, SDR theatre next, then SDR home and HDR theatre following, web stuff is not on the radar

for this i usualy get Dmin/Dmax wrapped inside DPX, use ACES, ADX10 IDT

a 709 source is limiting -> problematic -> rejected

 

[Giles Sherwood]

Since you expressed interest in using Filmbox for this, I would recommend just starting out with the built-in Resolve Kodak 709 LUT and then tweaking the overall look to taste before going in and doing your shot-to-shot corrections. IMO this tends to give a better starting point on film scans than the modern film emulation packages that are geared for digital sources. Personally, for film scans, I vastly prefer the results of a print LUT workflow to a fully manual one.

 

[Chris Wright]

I would definately ask more about the captured colorspace and gamma. If they already burned in Rec. 709 into your Log and Best Light, you're not going to lose anything working in Rec. 709(mainly greens and primary colors), but if you got a larger gamut captured, don't clip it if you don't need to.

 

[Juan Garassino]

I'm jumping into this conversation as a jr. Colorist that works with DVR because this is exactly what I'm researching.

I have ProRes 4444 XQ from a scann of a 16mm Bolex film (dont know the stock). The scan took place in some place in LA probably with a Scanity Scan according to their website, so I read the manual and on page 15 (dft-film) says:

File Format
10 bit LOG / LINEAR DPX according to SMPTE 268M-1994 under File Format.

My question would be: What should be my IDT? As basic as it is, I understand that a sensor bayern/codec light information into a codec that will be debayern/decodec this info into a wider gamut. So, in order for me to grade correctly I must know my input, hence Sensor's Color Science/Gamma, in this case its a scan, so its different and not so straight forward to know. Is Cineon Film Log the correct input? The image looks LOG enough to tell that no Rec709 has been burned.

 

[Rajneesh Kassin]

My question would be: What should be my IDT? As basic as it is, I understand that a sensor bayern/codec light information into a codec that will be debayern/decodec this info into a wider gamut. So, in order for me to grade correctly I must know my input, hence Sensor's Color Science/Gamma, in this case its a scan, so its different and not so straight forward to know. Is Cineon Film Log the correct input? The image looks LOG enough to tell that no Rec709 has been burned.

I deal with film scans all the time -- I think I just completed my 15th film feature for the year -- and the reality is they're so all over the place and unpredictable, I don't think a IDT could work. I just come up with an overall Printer Lights setting, followed by a Custom Contrast Curve, and then Balance and Gain nodes to get the image more or less under control. Subsequent nodes are used for matching and special issues, like lighting problems, flare, shading issues, and so on.

Brand-new negative, like Kodak 7219, can be extremely consistent and looks great under the right conditions. Older films are insanely inconsistent, especially when they're from decades ago, or from European or Asian sources, or (worse) patched together from multiple sources. Grading from film is tricky. A lot also depends on the accuracy of the people doing the scans.

 

[Chris Wright]

I'm jumping into this conversation as a jr. Colorist that works with DVR because this is exactly what I'm researching.

I have ProRes 4444 XQ from a scann of a 16mm Bolex film (dont know the stock). The scan took place in some place in LA probably with a Scanity Scan according to their website, so I read the manual and on page 15 (dft-film) says:


My question would be: What should be my IDT? As basic as it is, I understand that a sensor bayern/codec light information into a codec that will be debayern/decodec this info into a wider gamut. So, in order for me to grade correctly I must know my input, hence Sensor's Color Science/Gamma, in this case its a scan, so its different and not so straight forward to know. Is Cineon Film Log the correct input? The image looks LOG enough to tell that no Rec709 has been burned.

View attachment 13581

If you have a Prores 444XQ, its probably video levels, not full range. As for the gamma, it probably got encoded twice, once as LOG DPX/TIFF, and then again as Prores. Since they "burned" in the gamma from Prores, that's probably still LOG. But here's the catch, the re-encode probably took the Cineon full range capture and converted it to cineon video range gamma, and then burned that into Prores video levels. So, your IDT would actually be a video range Cineon, not full range!

 

[Chris Wright]

I deal with film scans all the time -- I think I just completed my 15th film feature for the year -- and the reality is they're so all over the place and unpredictable, I don't think a IDT could work. I just come up with an overall Printer Lights setting, followed by a Custom Contrast Curve, and then Balance and Gain nodes to get the image more or less under control. Subsequent nodes are used for matching and special issues, like lighting problems, flare, shading issues, and so on.

Brand-new negative, like Kodak 7219, can be extremely consistent and looks great under the right conditions. Older films are insanely inconsistent, especially when they're from decades ago, or from European or Asian sources, or (worse) patched together from multiple sources. Grading from film is tricky. A lot also depends on the accuracy of the people doing the scans.

I would recommend the grader understand the pro's and cons of using a custom contrast curve and I would enphasize that it should probably be used only for more seasoned colorists due to the many complications in handling the data such as disabling color management from broken or incorrectly setup scans.

 

[Juan Garassino]

Thanks, @Rajneesh Kassin , but I'm still wondering: wouldn't I then be grading in the material's native color space without applying any transform? That would mean grading directly in Rec.709/Cineon Film Log. If that's the case, I would need to normalize each clip manually using the pipeline you suggested.

@Chris Wright , I noticed that yesterday. However, when I changed the Clip Attributes in the Media Pool to Full Range, the image became more "washed out", compressed. The lows for example jump from 128 to 256 in a 10bit scale. That makes me think, is the codec actually using Video Range. Set it to Full Range when it's in Video shouldn't drop the lows instead of increase them?

So, my conclusion so far for a 16mm film scan (Bolex camera) is:

• If I don't apply an IDT, I would be grading directly in the film scan's native color space (Video Range/Full Range, depending on how the files are encoded).
• If I do apply an IDT, the pipeline would be something like: Rec.709 / Cineon Film Log → DWG/Intermediate → DRT JP2499. @Juan Pablo Zambrano FTW
• And if I work in ACEScct, then ADX10, as @Dermot Shane mentioned earlier, seems to be the appropriate choice.

 

[Rajneesh Kassin]

If you have a Prores 444XQ, its probably video levels, not full range. As for the gamma, it probably got encoded twice, once as LOG DPX/TIFF, and then again as Prores. Since they "burned" in the gamma from Prores, that's probably still LOG. But here's the catch, the re-encode probably took the Cineon full range capture and converted it to cineon video range gamma, and then burned that into Prores video levels. So, your IDT would actually be a video range Cineon, not full range!

No, all the 444XQ scans I get are from the ArriScanner XT, and that goes directly to ProRes 444XQ. No mid-format transcodes.

Thanks, @Rajneesh Kassin , but I'm still wondering: wouldn't I then be grading in the material's native color space without applying any transform? That would mean grading directly in Rec.709/Cineon Film Log. If that's the case, I would need to normalize each clip manually using the pipeline you suggested.

Don't make it more complicated than it needs to be. As far as I know, the color gamut from the scanners are still Rec709 -- it's the gamma curve that's tricky. I do keep a close eye on out-of-gamut colors, which can happen with police car lights, explosions, neon lights in nightclub interiors, stuff like that. Normalizing for me is all done in the first few nodes.

 

[Giles Sherwood]

Don't make it more complicated than it needs to be.

+1

 

[Chris Wright]

Thanks, @Rajneesh Kassin , but I'm still wondering: wouldn't I then be grading in the material's native color space without applying any transform? That would mean grading directly in Rec.709/Cineon Film Log. If that's the case, I would need to normalize each clip manually using the pipeline you suggested.

@Chris Wright , I noticed that yesterday. However, when I changed the Clip Attributes in the Media Pool to Full Range, the image became more "washed out", compressed. The lows for example jump from 128 to 256 in a 10bit scale. That makes me think, is the codec actually using Video Range. Set it to Full Range when it's in Video shouldn't drop the lows instead of increase them?

So, my conclusion so far for a 16mm film scan (Bolex camera) is:

• If I don't apply an IDT, I would be grading directly in the film scan's native color space (Video Range/Full Range, depending on how the files are encoded).
• If I do apply an IDT, the pipeline would be something like: Rec.709 / Cineon Film Log → DWG/Intermediate → DRT JP2499. @Juan Pablo Zambrano FTW
• And if I work in ACEScct, then ADX10, as @Dermot Shane mentioned earlier, seems to be the appropriate choice.

I'd rather use P3 than Rec. 709 just due to gamut size limitations. You usually don't run into them most of the time, but it's not guaranteed.
Some people even use D60 P3 for the IDT to get a one click grade. Now, if you are an Aces purest, the ADX10 will try to average out all film stocks together using its built in Cineon gamma. This sometimes works and sometimes doesn't, depending on how similar your film stock is to the early 2000's film stocks. Peter Ch used to enjoy P3 above the rest. As long as you don't clip anything, there's really no wrong answer. Just keep in mind the total workflow as many print luts require a certain type.

 

[Elias Nemer]

from what ive learned over the first few months/year is as such. Film primaries are not like digital/standardized, they are not like rec709 and such, maybe closer to p3, but i would do like @Rajneesh Kassin is suggesting and not to complicate things. the gamma is definatly the trickiest part, so drawing a nice s curve can put you in the right direction as a global correction plus a offset correction, and then work on clip basis, to normalize the footage and fix issues, as stated before.

that being said you could also use the kodak2383 resolve built in lut, as a print, its pretty good. you could if you wanted try to copy the s curve from there and then tweak to your liking. For the gammut you can have several options, but i dont think you would be too far off, you could hue some color add saturation in a nice way, whatever fits your project.

so generally IDT i dont think is neccessary as film is complete different from digital, in that the primaries are not consistant/standardized. each film is different and building you own transform i think works best and is the most ellegant solution.