Don’t worry, I’m working on sample videos for both regrades, and they should hopefully be up soon… 😃
That looks amazing! Glad to see the balancing tool is put to good use. 😃
The DVD has the best exposure of any Raiders transfer. Skies are clear and blue, nothing in the ark opening scene is overexposed. The wowow has some issues and the blu-ray is much worse. They should actually be better since they were created from a negative scan and the DVD from a 2nd gen print. Either the scan was not done correctly or (most likely) things were overexposed during grading.
You will definitely run into some challenges trying to match the wowow to the DVD (and LPP), particularly during the ark opening.
Yes, it will be interesting to see what details can be retrieved from the overexposed frames of the WOWOW.
This shot looks substantially different for the DVD, when compared to the WOWOW. To be honest, I think the WOWOW looks pretty awful for this shot.
DVD:
WOWOW:
WOWOW matched to DVD:
To take a break from the bar scene discussion, here’s a regrade for a completely different shot.
35mm LPP:
WOWOW:
WOWOW matched to 35mm LPP:
Here’s another regrade example with subtle, but noticeable differences in color timing, particulary skin tones, which tend to be more pink for the WOWOW, and more bronze for the DVD.
DVD:
WOWOW:
WOWOW matched to DVD:
I know this is probably a long shot, but any chance of pulling the snake reflection, rolling rock bar out of the LPP scan? i.e. to replace onscreen content that’s simply not present in the HDTV version?
I don’t think he wants to get into any roto or clean up, so he’s just using the WOWOW and BD.
I figured that was likely the case, but I was tempted to ask because it would be so great.
That would be beyond the scope of the project, but maybe someone else would like to take on this challenge?
It appears, that the bluray in some respects has more detail for this shot than the WOWOW, namely the skin and eyes, but not for the background, like the water and fog.
WOWOW matched to DVD:
Bluray matched to DVD:
LOL Built in fix! That’s great!
If the smoothing is levels-targeted, I may suggest twice of what you just did? 1x leaves some residue (viewed when angling this laptop’s LCD screen) but 2x should eradicate it.
I noticed the glare on Belloq’s forehead, so I’m guessing this problem is throughout the DVD. Also, I see edge enhancement there, too (thick darker/lighter blue sky edging the actors). I’m amazed that faux coloring didn’t carry over to your regrade. Built in fix, too?
Here’s the regrade with even stronger smoothing.
Regrade before:
Regrade now:
The faux coloring filter is built in, but not on purpose 😉.
Here’s another example. Notice, that the oranges look more like lemons for the WOWOW. Also, note the increased detail of the sand under Indy’s feet.
DVD:
WOWOW:
WOWOW matched to DVD:
This 1981 featurette has a short clip of the filming of the bar scene, and shows the lighting conditions used during filming are consistent with the colors of the LPP. The clip can be seen approximately at the 8:20 mark:
Excellent work! Automation and math is the way to go. ;D
The DVD has inflated highlights (at least in this shot). It carries over into your processing and looks bad there, too. I tried a deflation on the DVD with the lower end locked and non-singular R-G-B adjustments (to put some color back in). Not perfect, but it gets the artificial glare out. (Compare it to the more natural WOWOW.) If the entire DVD is this way inflated, a preliminary, single correction should produce a superior final result.
Actually, the issue with the glare can be minimized, by increasing the smoothing parameter of the algorithm, which reduces strong gradients.
Regrade before:
Regrade now:
Here’s another example. Although the WOWOW’s color timing is similar to the DVD, the DVD has warmer tones for this shot. The skin tones also tend to be less pink and more natural, than for the WOWOW.
DVD:
WOWOW:
WOWOW matched to DVD:
So for Raiders, there will be a regrade based on the 35mm print and a regrade based on the DVD?
Exactly!
This thread is dedicated to regrading the Indiana Jones trilogy, using the 2003 DVDs as a reference. Unlike the currently available HD releases/broadcasts, which were all sourced from a scan of the negative, the DVD was sourced from a color timed interpostive. Many consider the DVD color timing to be excellent, hence this project to reproduce these colors in high definition. While I will be using the bluray versions for Temple of Doom, and Last Crusade, I will be using the WOWOW broadcast for Raiders of the Lost Ark, as it is more detailed, and suffers less from DNR.
The regrade will be done, using the color matching tool I’ve developed, to ensure the colors are as close to the 35mm reference as is humanly possible:
Here’s a first regraded example frame from Raiders of the Lost Ark. Note the orange cast that pervades the color timing of the bluray.
DVD:
Bluray:
WOWOW:
WOWOW matched to DVD:
Here’s a comparison between the WOWOW and the regrade:
Here’s another regrade for the bar scene, such that everyone will have a better idea what it will look like.
35mm LPP:
WOWOW:
WOWOW matched to 35mm LPP:
Regarding the SOTS 35mm reference to 16mm target … I noticed that all matching/prediction attempts produced colorized tree-trunk shadows for the 16mm. (Such poor distributions are allot easier when viewed on the LCD screen I’m using, simply by changing my head position – never before thought of that as a plus.) I’ve come across something similar when using HSL color transferring between same pictures.
When the L (lightness) of the 2 pictures have differences (contrast, brightness), the transferred HS (hue,saturation) “colors” are altered by the destination’s different L, and sometimes unacceptably so. I’ve found that pre-matching the lightness of the destination to the source (very easy to do on the B&W L splits) produces a proper color transfer result.
Enough of your excellent matches show such anomalies that they cause you constant push-here-pull-there touch ups. Would some sort of R-G-B pre-normalization step, to create a better distribution for your limited samplings, save you such re-adjustments? (I would suggest using a larger sample pool, but the increased storage and processing requirements for such diminishing returns might be prohibitive.)
The latest version of the tool does better in this respect, but some anomalies are unavoidable. I will take a look at the method you’re suggesting, and see if I can incorporate it in some way.
What’s particularly weird for me in the bar scene on the print is you can clearly see their faces illuminated by red light, but there is nowhere in the frame that the warm light is originating from. The look that ended up on the print HAD to be deliberately put there during the grading of the film, because there is no way those were the colors on set. It’s actually perplexing me how they ended up with that look, as even if you have never seen Raiders before and have no preconceptions of the color, that would just look wrong and off putting, which makes me think it might have been a mistake, but one made too far along in the process and that didn’t seem a big enough deal to justify the price of fixing it.
It’s weird, it reminds me of some of the wacky experimental color grading done in the early 2000s when digital grading started to take off. The regraded WOWOW actually looks more like an 80s film. Just all around strange to me and I wonder how it ended up that way.
Actually, the bar itself emits red light in the print, so this is the source of the red light on their faces.
Understood, warts and all. He he, just kidding.
But I think digital colour adjustment to one scene could have been done without it affecting the entire movie. You just had to work on it separately. Though I get the idea of preserving the original print as is. Did you project the print, by the way?Also, was some scenes missing from the 35mm LPP?
Litemkr did all the work on the print scan, so he can probably be more helpful in this respect. 😉
The amount of correction you’re suggesting, would give all the daylight scenes a pronounced yellow/orange cast. Since all the other scenes have perfectly balanced colors, the colors for the bar scene should be accurate, as far as the original theatrical color timing is concerned. The fact that the same color timing can be found on an early home video release confirms this.
See, I am not asking you to do any of this, just expressing my honest opinion. There is no way that lantern was meant to give out a blue glow. It has to be yellow, it’s common sense (my tone is not rough or harsh, in case you read it as such). Just because all other scenes look ok doesn’t mean this shot cannot be wrong. Human error happens.
Then you would have to factor in the yellow-bulbs making the print more yellow. The DOP would light the shots and the colour timer would time them keeping in mind the warm projection bulbs, which would alter the col temp of the image.
Also, no need to alter the other scenes. They are more or less fine, colour-wise.
A VHS or laserdisc cannot be reliable sources of original colour timing.The VHS and laserdisc (and DVD/BD) of Jurassic Park, all had a blue colour-cast which was totally wrong for decades until the 4K remaster showed that the actual tinting was much warmer. In fact, 35mm frames from the prints floating in the internet confirmed this warmer timing. Though the 3D BD went slightly overboard with the orange.
I also remembered the Leaky Cauldron tavern scene in ‘Harry Potter & The Philosopher’s Stone.’ It was a similar low-lit, scene filled with candles and lanterns. And that scene also had a warm yellow-golden hue.
Yes, but here’s the problem with what you’re suggesting. The print’s colors are consistent between shots. Hence, you cannot adjust just one scene, while leaving the others unadjusted. So, you either end up with all the other scenes having a pronounced orange/yellow cast, or you accept the colors as they are, the bar scene included. It seems Spielberg at some point wasn’t happy with the look of this scene and felt like you did, that the lights did not look right amongst other things, and had the scene regraded. I might even agree, that the WOWOW color timing looks better, and more realistic for this scene. However, this is totally irrelevant to this project, as the goal is to reproduce the original theatrical color timing, including any color timing errors, that were in the original theatrical release, not the most realistic color timing.
This discussion is similar to the discussion on the theatrical color timing of The Empire Strikes Back, that originally had a pronounced blue cast for the Hoth scenes. Many then too felt that it looked “wrong” or unrealistic. However, ultimately it’s not about what we feel are the correct colors, but what a reliable color reference tells us the colors should be. The 4K remaster, like the bluray is revisionist. The film was completely retimed from the original negative. As such, it does not represent the original color timing, and there’s no reason to assume they even attempted to reproduce the original color timing. We have an original unfaded 1982 LPP, with the original color timing, there’s no more reliable color reference than that.
However, this thread is about regrading the WOWOW to what I and many others consider a reliable color reference. Discussion about the reliability of the 35mm LPP used should be directed to the appropriate thread, dedicated to this color reference:
Here’s an example for the bar scene.
35mm LPP:
WOWOW:
WOWOW matched to 35mm LPP:
This is what the unfaded print looks like, which is also consistent with the 35mm trailers from the era, and the screenshots of the early VHS release I posted. The warm bulb might add a little warmth, but in the same way that the warm setting on your tv will make it appear a little warmer. It will not make blue/white light appear yellow/orange like for the WOWOW. These are the colors of an unfaded print release print, as such it by definition cannot be grossly wrong.
Well, the best guess then is that Spielberg and the DOP got it wrong! The 30th anniversary 4K remaster was supervised by Spielberg, which is also warm tinted. Now, I agree that the red-push was a bit too much, but surely they looked at the original prints and know better than anybody else what is the right colour.
The best bet would be to catch a 35mm screening of Raiders in a movie theatre with '80s projection bulbs and the whole bunch.
Don’t get me wrong though. I seriously appreciate the time and effort you guys are putting at recreating an experience. I just think, in that particular scene, the lantern colour is totally incorrect.
The amount of correction you’re suggesting, would give all the daylight scenes a pronounced yellow/orange cast. Since all the other scenes have perfectly balanced colors, the colors for the bar scene should be accurate, as far as the original theatrical color timing is concerned. The fact that the same color timing can be found on an early home video release confirms this.
Here’s an example for the bar scene.
35mm LPP:
WOWOW:
WOWOW matched to 35mm LPP:
Forgive me, but an oil lantern does not glow blue, it glows with a yellow-golden hue. The 35mm colour cannot be correct. Surely Spielberg and Douglas Slocombe knew what a glowing lantern looks like. The WOWOW colour looks perfect. It’s the details that one must look into, I think, not just a source to grade properly. Also, In the 1980s and '90s the projection lamps used to be yellow, I think. So, any film projected would have actually looked warmer than how it appeared on the print itself. I am not an expert, but the 35mm colour looks grossly wrong and I do not think Spielberg wanted a yellow lamp to look blue in 1981.
P.S: Just felt a strong earthquake, shook the entire building. Whew, was a strong one. Our dog hid under the sofa.
This is what the unfaded print looks like, which is also consistent with the 35mm trailers from the era, and the screenshots of the early VHS release I posted. The warm bulb might add a little warmth, but in the same way that the warm setting on your tv will make it appear a little warmer. It will not make blue/white light appear yellow/orange like for the WOWOW. These are the colors of an unfaded print release print, as such it by definition cannot be grossly wrong.
Turisu said:
Conversely, the lamp that Indy and Marion are tied to in the opening of the ark sequence was digitally illuminated in the BD whereas it should be off. Not sure whether the WOWOW has the lamp on or off.Can I see some screenshots of this?
Also, how’s the matching tool do for the bar scene?
Here’s an example for the bar scene.
35mm LPP:
WOWOW:
WOWOW matched to 35mm LPP:
If you load the images in to photoshop, and sample sections of her jacket and their skin tones, you’ll see that the sampled pixels are more saturated. This is also confirmed by my eyes when I load the images in to Resolve and view them on my rec709 monitor.
Yes, but there’s a very good reason for this. You’re using the wrong metric. The print has much less detail in the shadow areas, when compared to the WOWOW, even taking into account the lower contrast of the WOWOW. A more detailed source will always have higher peaks in saturation and contrast (and lower lows). However, if you transform the colors to the HSV color space, you will observe, that the distribution for the saturation channel is the same for the regrade, as it is for the print. This is the relevant quantity to consider.
Per the crushed blacks; the release print is several generations away from the color-timed negatives, and will have more degeneration in these sensitive areas (more blown highlights, crushed blacks).
The saturation problem is relative to the print screenshots you’re posting; you’re increasing the contrast to match the print, but you’re not lowering the saturation to match, which, relative to the print picture you’re posting, results in a over saturated image. As such, the algorithm is simply not accounting for this.
The release print is what creators intended to be seen. The low contrast interpositive was not intended to be seen by anyone, but the creators.
With regards to the saturation, being the creator of the algorithm, I can assure you, that the algorithm matches hue, brightness, and saturation. The print is more grainy, which may make it seem the regrades are more saturated, but this is definitely not the case. It is a mathematical certainty. Here’s an example for another high saturation print scan, that I used for a test a while ago:
35 mm reference:
Bluray:
Bluray matched to 35 mm frame:
It’s pretty obvious, that the algorithm has matched hue, brightness, and saturation.
I suggest you read through the thread dedicated to color matching and prediction:
The colors look a bit off in the pencil scene. It looks like it should be graded to more of a teal color, as per the screener, but seems to appear more blue after it is run through DrDre’s CC tool. This causes some inaccuracies in flesh tones as well as costume colors.
Here’s a comparison between the original scene and a correction from the BD I performed to show you what I’m talking about:
http://screenshotcomparison.com/comparison/181853
The “table scope” part is referring to the rotoscope and minor CC of the table from the screener correction, as the table appears to have been digitally altered for the Blu-ray. This caused unnatural grain in areas, and is the likely cause of the fringing near the table’s upper edge.
To be honest, I think your correction, while appealing, is pretty far of the mark, if the intent is to match the colors of the screener.
Bluray:
Screener:
CC tool + black level correction:
zxthehedgehog:
