Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country: 
Latest comparison clip: https://vimeo.com/203985185
Top: LD untouchedw, BD color grading 2
Bot: BD color grading 1, BD untouched
I think I'll go with color grading 1, because it's closer to LD; however, in the eventual case of artefacts, I'll use the color grading 2 for that shot.
I also enhanced a bit the contrast for color grading 1, as well as saturation increase (maybe too much?), to stay even closer to LD.
Opinions?
Posts: 1,186
Threads: 54
Joined: 2015 Jan
Thanks: 96
Given 98 thank(s) in 80 post(s)
Country: 
I have the U.S. Collector's Edition DVD. Would there be any use for anything on it for you?
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
I think it contains the same material found on the PAL DVD; thanks for the offer, if it will be a need for it, now I know you have it.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
Update: right now, the first pass encoding is on its way, thanks to the new computer; I have chosen this project just because it's the only one almost finished and ready on an external hard disks, and I had to test the PC, so...
I've decided to go with the old color matching script; even using this PC, encoding now is expected to be finished in 7hrs, while using the new combined one it should take three to five times more and, even if artefacts are almost non-existent, color fidelity is lower; so, I'm expecting few artefacts here and there, but probably will take few times to correct those, using the new combined script just for the eventual bad shots, than make the whole with it; plus, colors will be closer to the reference.
Tomorrow I'll be away, so I expect to work on this next week.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
Got a "we ran out of memory, folks!" error message after less than 20000 frames... 
Now I put the setmemory at the beginning, let's see if everything goes well.
Posts: 2,290
Threads: 39
Joined: 2015 Jan
Thanks: 177
Given 187 thank(s) in 133 post(s)
These are just "growing pains" of getting to know your new equipment 
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
Well, this time the error appeared just after 500 frames...
I wonder if it's due avisynth version (difficult), missing filters/script (quite difficult), or the fact scriptclip adds up memory - until virtualdub eventually crashed (that's the most probable cause).
Now I just figure out how to turn the scriptclip "normal" part to the one with no memory problems... but I'm too tired now...
Posts: 3,291
Threads: 176
Joined: 2015 Jan
Thanks: 1450
Given 977 thank(s) in 494 post(s)
Country:
Sorry to hear that, very frustrating indeed.
Film Addict
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
Yep, really... I spent the whole day yesterday to find a workaround, and finally found it!
In small tests, it works perfectly - apart the usual artefacts...
About this, I could decide to follow one of two paths:
1) very close color reproduction (IMHO 95% and up), but some artefacts on clipping whites
2) lesser color fidelity (IMHO 80% up to 90%), with no artefacts (until now)
For now, I stick with 1) and then replace the shots ridden with artefacts with 2) (or just affected parts, going to investigate)...
If everything goes well, final encoding will take about 7 hours, not bad in comparison to days!
EDIT: meanwhile, I reviewed the crashed version; so far, more than 25000 frames were correctly encoded (that's about 1/6th of the movie); on them, I spotted four instances of artefacts; one was really noticeable, and last few dozen seconds; the other three were from almost unnoticeable to barely noticeable, and last few frames each. Not the best news, I must admit, but neither the worst; it will take relatively few times to replace them with the second version, so I think at the end there will be around 98% of the footage with great similarity with the color reference - as I wrote, I bet the worst of them would retain around 95% of the color fidelity, while many would be almost identical - always taking in account the fact LD was captured with low contrast and saturation, and those aleatory figures could be used in comparison with corrected LD; the residual 2% should still be rendered, but I expect a minimal of 80% and more - during tests, most of the shots were really close, while few have minor differences, with some colors inherited from the BD in the second plugin, while the first (almost) perfectly replicated the LD colors.
|
