We'll see. Right now our main problem is that we might need to re-render all three movies to get rid of the red bleeding. Some analyzing showed it wasn't caused by the encoder, but happened when rendering the avisynth script. And if a re-render is needed, it means we'll need to redo all the bad frames again I guess.
Is the sample on myspleen or on a torrent site?
Sorry I've been lurking for quite some time and finally to just join since I'm a huge fan of the work youguys have been putting into this project.
adywan said:
It's such a sad state of affairs when people with a PC at home can do a much better job than the half assed effort of Lucasfilm. Its sad to think that there are so many Star Wars fans that truly think that the 2004 DVD/ Blu-ray versions are how Star Wars should look colourwise. These fans should be forced to watch this version as THIS is how Star Wars should look.
Great job guys.
I think you hit the nail on the head with that one Adywan. My thoughts exactly. You_Too and DJ, simply amazing work; you guys are heroes. But really.
May the Force be with you.
You_Too said:
@patkhoo: What I did to find the balance in the movies was to take lots of frames from each movie, put them in photoshop beside each other in one big file, and let photoshop auto find the mid-balance of the greyscale.
Also, keep in mind that the movies were projected with warmer bulbs back in the day, as Mike Verta discussed in Harmy's thread. This would make them look more warm than the 70mm scans, which I bet were scanned using today's standards.
And this shot was just one of very many in that movie. Some shots still have a very cold feeling to them. The GOUT does not do a good job in representing the original color but I think we did the best we could to bring some of it back.
Ah, Gotcha! Were you using auto on levels or curves? Or both? I understand putting many frames into one big image, but would that be equivalent to finding the average of 10 when 3 are more like "X", 2 are closer to "Y" and the last 5 closer to "J"? I'm sure you considered this already, so what was the conclusion you came to? Related is, how was the previous (DVDv3) methods used (in comparison to this new method), which incidentally is clearly nowhere as good as this new method..
I'm not criticising the new approach, just trying to understand the rationale behind the methods used.
And now, for some fantasy.. I wonder.. (just wondering only)
The "auto-" capabilities in Lightroom are simpler to use than those in Photoshop. It is conceivable (damn near impossible with Photoshop, but maybe possible in Lightroom) to extract every single frame of the movie and adjust white balance, tint, saturation, vibrancy etc, for a "typical frame" in a schene, then apply the same settings in batch for all frames of the same scene. "Copy Metadata" and "Import profiles" are the key techs. Then Lightroom will batch render all frames to be recombined into a lossless AVI later. The fantastic thing is of course the fact that Lighhtroom does not make any changes to the actual image itself, only saving it in the DB, it is therefore possible to make changes to an individual scene (as opposed to a frame) as and when required.
In addition, the dirt removal tool is point and click for problem frames etc and advanced tools like unsharp mask, upscaling to 1080 etc are all a preset away to be applied on Lightroom's "Export" function. Heck, even the "dirt removal" tool can be batch applied to multiple shots in Lightroom..
Of course, let's see, a 120 min movie @ 30 frames/s = 12,960,000 frames. If there are 1000 scenes in a movie, then just manually adjust 1000 sample frames, and then batch apply to the remainder.. OK, it is fairly difficult, but not impossible. This may be the nearest practical method to the ultimate unthinkable of manually adjusting each and every frame in Photoshop...
Let me just say, it is merely a thought excercise, but I have Lightroom (and Photoshop), and I wonder ...
patkhoo said:
Of course, let's see, a 120 min movie @ 30 frames/s = 12,960,000 frames.
how did you get that?
2 hr = 2 hr x 60 min/hr = 120 min
120 min x 60 sec/min = 7200 sec
7200 sec x 30 frames/sec = 216000 frames
(if it was the normal 24 frame/sec) = 7200 sec x 24 frames/sec=172280 frames
i started a frame reference count for the whole
movie in the technical forum:
------------------------------------------------------------
http://originaltrilogy.com/forum/topic.cfm/frame-data--star-wars-GOUT-NTSC-widescreen/topic/9567/
only got through the first reel so far, around 27000+ frames.
later
-1
[no GOUT in CED?-> GOUT CED]
patkhoo said:
Ah, Gotcha! Were you using auto on levels or curves? Or both? I understand putting many frames into one big image, but would that be equivalent to finding the average of 10 when 3 are more like "X", 2 are closer to "Y" and the last 5 closer to "J"? I'm sure you considered this already, so what was the conclusion you came to? Related is, how was the previous (DVDv3) methods used (in comparison to this new method), which incidentally is clearly nowhere as good as this new method..
I'm not criticising the new approach, just trying to understand the rationale behind the methods used.
I used auto on curves, with zero on clipping of course. Of course it wouldn't be equivalent of the perfect average, but on the other hand, it will almost always find the most obvious tint and fix it.
As for DVD V3 I don't know since I wasn't involved back then.
And your idea with Lightroom is a very good one, but we won't go for such an advanced approach. We'll keep our color settings as they are and work from there. Feel free to start your own project! :)
negative1 said:
7200 sec x 30 frames/sec = 216000 frames
What The ... :D
More proof why my maths teachers hated me in school.. sigh! :D I now have no idea how I got 12 million.. I stand corrected.. but it in fact makes Lightroom idea more possible. Only 216,000 "photographs" to correct.. :D
You_Too said:
patkhoo said:
Ah, Gotcha! Were you using auto on levels or curves? Or both? I understand putting many frames into one big image, but would that be equivalent to finding the average of 10 when 3 are more like "X", 2 are closer to "Y" and the last 5 closer to "J"? I'm sure you considered this already, so what was the conclusion you came to? Related is, how was the previous (DVDv3) methods used (in comparison to this new method), which incidentally is clearly nowhere as good as this new method..
I'm not criticising the new approach, just trying to understand the rationale behind the methods used.
I used auto on curves, with zero on clipping of course. Of course it wouldn't be equivalent of the perfect average, but on the other hand, it will almost always find the most obvious tint and fix it.
As for DVD V3 I don't know since I wasn't involved back then.
And your idea with Lightroom is a very good one, but we won't go for such an advanced approach. We'll keep our color settings as they are and work from there. Feel free to start your own project! :)
Ah, understood.. Auto-curves would work pretty well.. Yes, you're right, the obvious tint faults for the entire movie would have been resolved in this manner - in a single step, no less. Excellent creative thinking - my salute to you, You_Too! (hey a palindrome.. I digress..)
As for the Lightroom idea, well, I don't have access to any files, videos, images, etc, so there's no chance for me to do such a thing - like I said, a bit of a thought experiment is all.. although if someone is keen to give it a try, I can contribute help on the Lightroom and imaging front.. A PM is best.
Anyway, back to our regularly scheduled program - Blu V2, when I can have it..??? OK, OK, all together now, "It'll be ready, when it's ready.." :D
Well, I watched the Blu V2 teaser a couple times now. I realize I'm likely in the minority thinking this, but I find the image overly grainy/noisy. Compared with the V3 DVD, the picture actually looks less clear in the Blu version. I know an attempt was made to preserve as much detail as possible, and that no one is a fan of waxy-looking surfaces as a result of too much denoising, but I feel the picture quality could benefit from a little 'cleaning up' or 'smoothing out'. It feels at times that I'm watching the movie through a filter (if that makes any sense).
My apologies for sounding critical. I've been following this project for a long time and am very appreciative of the tremendous effort that has gone into this preservation. I just wanted to know if the image quality of the teaser will be reflective of the final product and, if so, if the level of grain/noise (sorry if I'm not using the correct terminology here) is likely to remain as is.
Just wanted to know what DJ's and You_Too's -- and others' -- thoughts were on this issue.
“It’s a lot of fun… it’s a lot of fun to watch Star Wars.” – Bill Moyers
corellian77 said:
Well, I watched the Blu V2 teaser a couple times now. I realize I'm likely in the minority thinking this, but I find the image overly grainy/noisy. Compared with the V3 DVD, the picture actually looks less clear in the Blu version. I know an attempt was made to preserve as much detail as possible, and that no one is a fan of waxy-looking surfaces as a result of too much denoising, but I feel the picture quality could benefit from a little 'cleaning up' or 'smoothing out'. It feels at times that I'm watching the movie through a filter (if that makes any sense).
My apologies for sounding critical. I've been following this project for a long time and am very appreciative of the tremendous effort that has gone into this preservation. I just wanted to know if the image quality of the teaser will be reflective of the final product and, if so, if the level of grain/noise (sorry if I'm not using the correct terminology here) is likely to remain as is.
Just wanted to know what DJ's and You_Too's -- and others' -- thoughts were on this issue.
I brought this exact same thing up myself, I thought it looked more grainy than ever before lol, but U2(short for You_Too lol) said this is how it looks and it brings out detail, when I first started all these tests 3 things stuck out to me right from the bat,
1 was the color, I was so used to it being so dull and washed out, no life, that the new color threw me off a bit, but I love it now!
2 what you mention, the grain seems really heavy, a lot heavier than I have ever seen before on any GOUT project, now I am not saying I do not like it, but this will take some getting used to.
3 the occasional flickering is what I will call it(I am not real good myself on all the real tech terms lol) like during the pan down, the flickering on the walls and ground, and most of all, the flickering on the foliage behind our Heroes at the end of the film.
But for right now things are delayed a bit, we have to fix a tech issue we found mentioned above(sure wish we would have caught it sooner lol) so we are running a ton of tests before we render a new final lossless avi file, because this next file will be final, so we will look at some of the concerns brought up in this thread, like Ady's thoughts on the red, we can look into the "grain" issue, and if anyone has anything else to add now is the time, so please do so.
So does anyone else have any concerns with the grain share your thoughts now please.
Thanks, now it is time to get back to work lol.
The grain exists in the source, and to be able to dig up more detail I had to sharpen the films on the smallest level of fine detail. This brings up more details, it really does, but also sharpens some of the grain but like I said, the grain IS there already. If we smooth out the films it will look waxy and ugly because of the low resolution of the source, and there would be no reason to even make it a blu-ray, because a DVD would be more than capable of reproducing the detail that is left.
The flicker DJ is mentioning is also there in the source, and is a weird byproduct of the bad aliasing/interlacing and maybe also the noise reduction LFL applied to the prints which also gave us the smearing. The reason it's more visible is because we've stabilized the gate weave, thus making our eyes focus more on details than the overall wobble of the picture. It's also more visible since the image is more detailed now!
To show you a comparison between "smoothed out" and "detailed" I took one shot from moth3r's website which is from DJ's V3, and on the bottom is the Blu V2 downscaled:
The wall panels look washed out in the bottom pic.
I for one want the grain left exactly as it is in the sample. To my eyes it looks very sharp and looks more like a film print. Please do not DNR this release.
dlbsyst said:
I for one want the grain left exactly as it is in the sample. To my eyes it looks very sharp and looks more like a film print. Please do not DNR this release.
Don't worry, this will not happen.
Very cool!
The wall panels look f__king awesome in the bottom pic.
I love everybody. Lets all smoke some reefer and chill. Hug and kisses for everybody.
dlbsyst said:
I for one want the grain left exactly as it is in the sample. To my eyes it looks very sharp and looks more like a film print. Please do not DNR this release.
We won't use DNR. Of course there will always be different opinions and this version won't be for everyone, but rest assured that both me and DJ want to keep as much detail as possible.
vbangle said:
The wall panels look f__king awesome in the bottom pic.
Yeah, can you imagine all that color was actually there in the GOUT? :)
Once we got rid of the tint and desaturation, the true colors start to show up!
vbangle said:
The wall panels look f__king awesome in the bottom pic.
Yea I agree, I do not understand the "washed out" comment though, they look like nice white corridors to me, but to each their own right?
sample is coming down very soon.
What I mean is that the wall panels that stick out have a low contrast to the wall it self. As so the wall panels that stick out look flatter than they should as I thing some of the shadows of the panels was removed It could Just be my lap top screen though. But here is the bottom pic slightly altered http://i174.photobucket.com/albums/w83/Stitchfan_82/3poc.jpg
This has also been explained before, I think.
The movies have a very dynamic picture. I worked for weeks on it to make stuff like people's faces come out with more contrast like in the theatrical prints, and to be able to keep the whole range of contrast and still keep it balanced, I had to push the whites together a bit, without crushing them of course. The other option would be to crush the blacks.
I've kept the whole range and made the best of it that was possible. And I think some original prints were quite bright too, so I'm happy with this result.
Also remember I'm working on a calibrated screen. I think on a laptop screen you might even see crushed whites, making them look worse than they actually are.
crushed whites That’ was the term I was looking for. The 2004 DVD had crushed Blacks And on my laptop screen I have crushed whites. (“Also remember I'm working on a calibrated screen.”) Makes me think of something i was going to ask Harmy What is the chants that you may include some thing like a THX optimizer , so People can better Calibrate there screen for this version?
A THX optimizer is not calibration, it's basic setup.
And you can find such discs all over the web anyway, to be able to set the brightness, contrast and saturation/color of your screen.
Calibration is when you use a colorimeter to set the greyscale and each color to match a reference.
OK! There is a lot I do not Know. For one I thought that THX optimizer was specific to the movie it was on. When I would watch the GOUT I would use the GOUT THX optimizer if I watch the 2004 DVD I would use the 2004 DVD THX optimizer. Are all the optimizer’s the same?
I know this has been done to death on the other thread (and I've lightly touched on it before), but since you seem to still use some avisynth in this project and there is concern about red tinted faces...
AviSynth's built in Tweak filter: http://avisynth.org/mediawiki/Tweak has the ability for you to isolate a color (say the reds) and then adjust the saturation for just portions of it you want.
For example:
Tweak (StartHue=80, EndHue=135, MaxSat=80, MinSat=10, Sat=1.20)
These are just numbers out of my butt, but what it does is this:
Start/EndHue selects the reds (80-135) (you can narrow the range if some reds are weaker than others).
Max/MinSat selects only reds that already have a saturation between 10% and 80% of maximum.
This allows you to prevent low level reds that might be errors from being boosted or reds that are already strong enough from being oversaturated.
Now just reds between 10-80% in strength are increase by 20%.
Probably some trial and error needed, but it should let you give the red levels a strong boost, but avoid oversaturation skin tones.
Just a thought. (My inability to demonstrate this further is from my lack of reference sources, and mild color blindness.)
Dr. M
red5-626 said:
OK! There is a lot I do not Know.
For one I thought that THX optimizer was specific to the movie it was on. When I would watch the GOUT I would use the GOUT THX optimizer if I watch the 2004 DVD I would use the 2004 DVD THX optimizer. Are all the optimizer’s the same?
No, your first statement is correct. All the optimizers should be different to some degree.
The mind set behind THX is that the optimizer is passed through the same transfer process to the disc as the movie is. As a result any variation that occurs in the film will similarly happen to the Optimode allowing you to compensate.
Whether this is still true or not is anyone's guess, but it was at its inception.
And I don't know anyone crazy enough to recalibrate their TV every time they put a different movie on.
Dr. M
