[In progress] Hellraiser (1987) [Project: Celluloid]

[spoRv]

Say "thanks" to him from me, and our community (I guess)!

[deleted user]

I have an idea. With so many (DVD and Blu) sources available, you could try each and every one and try to find one that is consistent across all the shots after one pass through DrDre with all the images - indicating that it's consistent to the theatrical print too. If such a source exists, you could then use it for regrading the entire Arrow Blu Ray shot by shot, as it would basically become a perfect color reference after having the LUT by DrDre applied to it.

Of course it's likely a lot of work, but I think it's not entirely unrealistic.

The fastest approach to doing this (imo) would be to sync and align (as best as possible) all the sources in AviSynth, then load up in VirtualDub, go to the frame you need for the reference, and then just go "File->Export->Single Image...", then switch the source in Virtualdub, go "File->Reopen Video file" and take the same screenshot from the next source.

Or, you could even stack them all in AviSynth and just always save an image that contains all the sources, then go through all them afterwards (for example with a Photoshop automation) and split them up to individual files. Should be easy enough to do I think.

That way, you only have to search each corresponding frame once.

[spoRv]

I have an idea. With so many (DVD and Blu) sources available, you could try each and every one and try to find one that is consistent across all the shots after one pass through DrDre with all the images - indicating that it's consistent to the theatrical print too.

Or... median them (I mean, the regraded versions), and see what happens!

[deleted user]

I have an idea. With so many (DVD and Blu) sources available, you could try each and every one and try to find one that is consistent across all the shots after one pass through DrDre with all the images - indicating that it's consistent to the theatrical print too.

Or... median them (I mean, the regraded versions), and see what happens!

Oh yeah that too. Surely it's not too much work to sync up all the sources pixel-perfectly.

[LucasGodzilla]

I have an idea. With so many (DVD and Blu) sources available, you could try each and every one and try to find one that is consistent across all the shots after one pass through DrDre with all the images - indicating that it's consistent to the theatrical print too. If such a source exists, you could then use it for regrading the entire Arrow Blu Ray shot by shot, as it would basically become a perfect color reference after having the LUT by DrDre applied to it.

Well unfortunately from what I can tell, there are only two sources that are worthy of the consistency-to-print title: The Polish DVD (which I don't even have and it has the problem of looking wrong in the hospital-cenobite confrontation) and the laserdisc rip Sporv lent me (which has it almost dead-on in the hospital-cenobite scene but not anywhere else). Simply put, I don't have many sources of the movie that I could use, and even if I did, I don't entirely think they'd be useful.

My current plan is already essentially a shot by shot grade. Currently, I still have to make a grade for all of the new images and several subsequent merged grades for certain scenes. For example, in the specific shots where I have a still counterpart, I'll use that specifically made grade, but in other shots in the same setting (EG: The attic), I will use a grade made from all of the stills in that setting. For areas outside of that specific area, I will just a compilation grade again (only this time without the cenobite stills to retain relative, non-washed-out normality). Hopefully, that sounds fine to everyone.

Also, this is going to make me sound stupid probably, not that I am not since this whole venture is new to me, what do you mean by "one-pass"? Actually, let me rephrase that: How would you "double-pass"?

On that note, I'd also like to say that, by all means, y'all can experiments with your own with this movie and its theatrical print color grade. I don't want to be that prick who's hogging a movie despite minimal experience so if you think you'd do a better job at this then go right ahead, after all, I linked the gallery of the extra scanned stills to that pic in my last post if y'all didn't realize it already.

Oh, now that I remembered, I also want to ask something about Dr. Dre's coloring tool: How forgiving is it with scratches and whatnot (take this still as an example of something I'm worrying about). I'm also wondering what configuration should be used for the best results. Right now I set it to the max color space (250) with 0.1 smoothing. I feel like the color space thing is excessive but at the same time, idk: Is there a theoretical situation where maxing it out is bad?

[spoRv]

I guess that Tom intended, with one pass: "passed through" DrDre's tool (right, Tom?)

Hint #1: for the source(s) you use as color reference, degrain and denoise them; it will lead to a better result using any color matching tool.
Hint #2: why don't you use Polish DVD as main source, and the LD for the hospital scene? Because it will be easier to use two sources, instead regrading each scene by hand. Or, at least, use them, AND finally make the color correction you think are needed - it will take less time, I think.

[deleted user]

Well, hold on, by consistent I don't mean that it actually looks like the print. Distinguish theatrically accurate and theatrically consistent.

What I meant is that the difference between colors from scene to scene *relatively* is consistent to how it's in the print. So in other words, if you colormatched say 20 stills across the entire movie with DrDre as a single LUT, a consistent source would give the desired result in each single one of the shots and as a result likely also in the entire movie. This would/could for example be the case with a scan straight from the IP that hasn't gotten any additional digital grading - it may not look right straight out of the box, but it can be adjusted perfectly with a single global LUT. Compare this to a digitally shot-by-shot regraded source, which would result in some scenes being a good fit and others not at all.

A good example of a not theatrically consistent source would be the Blade Runner UHD and the remastered Silence of the Lambs. When I attempted to colormatch them with a global LUT, some scenes for example became too bright/green while others became too dark/magenta as the LUT attempted to push them in the right direction globally.

I have a good example of a threatrically consistent source too, but I'm not revealing that yet, as I'm still working on it.

My idea was simply that with so many sources, there's a fair chance that one of the sources may be consistent in that manner.

In my experience, at least when dealing with multiple images in one, DrDre is extremely forgiving of noise/grain, as long as the images are well aligned. Bad alignment can definitely lead to problems though. After all, it calculates averages for what it's doing, so noise would naturally get lost. I never bothered denoising personally.

As spoRv said, I meant simply passed through DrDre, no focus on the "one". Though, coincidentally, sometimes 2 passes can actually help. For example with badly aligned sources, the produced LUT often has artifacts in the bright areas. This can often be improved by applying the LUT to the footage-to-be-regraded and then using that as a reference in a second pass.

Now, I'm not promising to do any major stuff myself, as I tend to be lazy, but would you perhaps share the pictures you got of the print? Maybe I'd play around a little, for my own amusement and maybe some lazy little project. As mentioned, I'm still holding out on that French Blu Ray, to see if it also has the same master, which would mean that I can use a Median between 3 sources of the same master.

[LucasGodzilla]

What I meant is that the difference between colors from scene to scene *relatively* is consistent to how it's in the print. So in other words, if you colormatched say 20 stills across the entire movie with DrDre as a single LUT, a consistent source would give the desired result in each single one of the shots and as a result likely also in the entire movie. This would/could for example be the case with a scan straight from the IP that hasn't gotten any additional digital grading - it may not look right straight out of the box, but it can be adjusted perfectly with a single global LUT.  Compare this to a digitally shot-by-shot regraded source, which would result in some scenes being a good fit and others not at all.

As spoRv said, I meant simply passed through DrDre, no focus on the "one". Though, coincidentally, sometimes 2 passes can actually help. For example with badly aligned sources, the produced LUT often has artifacts in the bright areas. This can often be improved by applying the LUT to the footage-to-be-regraded and then using that as a reference in a second pass.

Now, I'm not promising to do any major stuff myself, as I tend to be lazy, but would you perhaps share the pictures you got of the print? Maybe I'd play around a little, for my own amusement and maybe some lazy little project. As mentioned, I'm still holding out on that French Blu Ray, to see if it also has the same master, which would mean that I can use a Median between 3 sources of the same master.

Ah okay, that made a whole lot of sense. One thing I noticed was that it seems we are on the same page in terms of the color-matching numerous stills into a single LUT. This is what I referred to repeatedly as a Compilation LUT for a lack of better terms. Anyways, I did that and it led to some off-colorations at times likely due to the aforementioned cenobite confrontation where it seems to be quite different from the rest of the movie. Plan is to do a take two of that with the new stills and without the two cenobite stills I have and use this "Global / Compilation" grade for segments where I have no reference of (such as the lower levels of the house in the middle of the day or the pet shop and etc).

Your explanation of the double-pass bit also helped me understand what you meant by that. I may consider doing that but I will admit, finding all of the frames again (since VLC doesn't label any time nor frame stamps in the screenshots) is not going to be fun. It's certainly not out of the options though.

Anyways, sent you a PM with the specific links.

[LucasGodzilla]

Although this isn't exactly a project update, I just thought it'd be cool to share...

I don't know why it only occurred to me just now to look this up, but I realized that one thing that would serve as a good visual reference when grading the scenes without any scans, would be to simply look up film cells of the movie and use those as a guide.

Now obviously I'm not going to use these pictures in the Dr. Dre tool (or at least not without some editing), however, these pictures help me figure out what the general color scheme / temperature in some scenes without a reference as well as confirming the colors seen in JohnCarpenterFan's scans. The most helpful image I've found actually in the posted pics here is the one with the hobo in the pet shop since I wasn't ever sure how warm that part was until now. Similarly, I can say the same for the House-Cenobites scenes with Frank spinning upside down since I wasn't sure how bright those scenes were intended to be considering the fact how the Hospital-Cenobites scene looks in comparison.

[LucasGodzilla]

So uh, good news and bad news...

The good news: I have made nearly all of the LUTS necessary for the project and it's really just a matter of applying them now. The bad news: Da Vinci Resolve is going haywire (weird banding and "Missing Lookup Table" errors) and I can't determine the source of the problem apart from simply that my computer's hardware is too weak or something (which I call BS). It's so bizarre since when I apply a LUT that is "missing", it simply just doesn't do anything and for some of the ones that work, it does weird sh!t like this.

I even decided to check the LUT by applying it to the same scene with Dr. Dre's tool to see if it was just the LUT itself but it came out fine...

So honestly, idk at this point.