(2021-01-28, 06:13 PM)pipefan413 Wrote: the "uncalibrated" image is clearly extremely saturated and warm

It looks pretty dull and brownish to me. It seems like you're talking about the new scan, but he was referring to the old scan.

(2021-01-28, 08:54 PM)Dek Rollins Wrote:

(2021-01-28, 06:13 PM)pipefan413 Wrote: the "uncalibrated" image is clearly extremely saturated and warm

It looks pretty dull and brownish to me. It seems like you're talking about the new scan, but he was referring to the old scan.

Oh right, that makes a lot more sense. Yeah, I thought he was saying the *new* scan was uncalibrated and wondered how the hell it looked so much worse after doing said calibration. ("How the hell did all the scratches come back?!") Hahah. Sorry.

I was very much not saying the old scan looked good as is, I was saying the new one looked good and I didn't really understand why it'd need to be altered. But that's cleared that up, seeing as I now see (re-reading it after you said this) that it wasn't the new one he was talking about.

(2021-01-28, 06:13 PM)pipefan413 Wrote:

(2021-01-28, 07:15 AM)TomArrow Wrote: The comparison is a tiny bit unfair tho, given that one of the images is horribly compressed and not calibrated. Just adding calibration makes it look a lot different:

I apologise for what I'm sure is very much a rookie question but what specifically did you do to "calibrate" that second image? My assumption would have been that calibration would be something done before capture, rather than after, although I expect this is just a specific term referring to correcting white balance, levels, etc. Not to derail, but I figure others might potentially benefit from any response you might be able to give, whereas PM could only be of use to me individually. I'm mostly wondering how one would determine what it's actually meant to look like when doing this; is there something buried in Resolve (e.g. a LUT) that is theoretically right for various film stocks / colour systems?

Part of why I ask is that the "uncalibrated" image is clearly extremely saturated and warm but... it also kinda looks good as is to me? In isolation, at least; it might not necessarily be true of other shots scanned with the same settings.

In this case, I worked with the same scanner on another project and he scanned an IT-8 target for me. From that I was able to create a color calibration matrix. I simply applied that to this image after converting it to linear sRGB space.

A color calibration matrix is basically a set of 9 or 12 values that you can apply using, for example, the channel mixer in various software. Mine typically has 9 values. 9 values are basically just coefficients to multiplying the base RGB values. 12 values would also include a constant offset for each channel.

For example, a matix could be something like:
1.5 -0.5 -0.5
-0.5 1.5 -0.5
-0.5 -0.5 1.5

The output R value would then be: 1.5*inputR + -0.5*inputG + -0.5*inputB, and so on.

There is limited use trying to figure out the correct values by hand, unless you're fine with approximations. There are trillions of possible combinations. That's why these IT-8 (or other) targets are very useful.

(2021-01-29, 04:17 AM)TomArrow Wrote: In this case, I worked with the same scanner on another project and he scanned an IT-8 target for me. From that I was able to create a color calibration matrix. I simply applied that to this image after converting it to linear sRGB space.

A color calibration matrix is basically a set of 9 or 12 values that you can apply using, for example, the channel mixer in various software. Mine typically has 9 values. 9 values are basically just coefficients to multiplying the base RGB values. 12 values would also include a constant offset for each channel.

For example, a matix could be something like:
1.5 -0.5 -0.5
-0.5 1.5 -0.5
-0.5 -0.5 1.5

The output R value would then be: 1.5*inputR + -0.5*inputG + -0.5*inputB, and so on.

There is limited use trying to figure out the correct values by hand, unless you're fine with approximations. There are trillions of possible combinations. That's why these IT-8 (or other) targets are very useful.

Thanks, I'm with you now. (I misunderstood, I thought you were saying the *new* scan was uncalibrated and had adjusted that. But I'm guessing the new scan probably also is uncalibrated.)

I wonder if that IT-8 target (or a similar one) is still available to the scanner? Could be useful to check whether the new Cintel could do with post-scan tweaking in the same way.

Ah, I see. Pretty sure the Cintel comes factory-calibrated. Raw sensor data wouldn't look this saturated unless it's some special kind of sensor I haven't heard of.

Think of it like this - when you take a picture with any camera/DSLR or whatever, the true raw colors of the sensor look very desaturated, like that uncalibrated image. But the camera never shows you that raw data. When you take a JPG with a camera, the factory calibration is already burned into the data, aka the matrix is already applied (and other adjustments ofc). When you take a raw picture, it does save the raw data into the raw file, however it also saves the calibration matrix straight into the raw file and when you open the raw file in Adobe Camera Raw or similar, you already see the factory calibrated colors.

Afaik the scanner still has the IT-8 target, I basically gifted it to him. However I can imagine that with Cintel being specialized in film scanning (other than the typical DIY project industrial cameras), it might come with calibrations and automatisms specifically for each film type that might be even superior to the IT-8 target. I could be wrong about that, but I'd kinda expect that. Since the specific dyes used and the base color of the film can vary a little from stock to stock. This particular IT-8 target is really for Fuji slide film, but it gives pretty good results with any slide film except Kodachrome, from what I've seen. Kodachrome is a bit special since it's an entirely different process and needs a different target. I'm not knowledgeable much about motion picture processes, but the IT-8 target gave good enough results with everything I've seen so far. The Cintel people likely know more than me about that and with such knowledge they could perfect it further. Whether they actually did, I don't know.

Happy to put a bit more in, trying to remember - did you PM the link before?

Hi I have literally ended up here and even signed up chasing this Hellraiser re-scan! Can someone explain to a complete noob how I can donate and how I will be able to watch a 4K version of this when it's done? I have no way of burning 4K discs myself.....but I really want to donate and be able to see this! Looks fantastic

Are donations still being accepted for this? I'd love to pitch in.