Dufusyte

Mielr said:

sort of 'outlining' on the faces

If you have Sharpening cranked up on your display (tv) it can create rings and halos around high contrast objects.  I'm suggesting this might be caused by a setting on your hardware.  Try setting Sharpening = 0 on your tv.

The top shot looks like Luke is at the beach on a sunny day.  The bottom (v3) shot looks Luke is on a cold freezing stormy tundra.

Evidently the v3 hits the mark.  The original SW filmmakers may have color corrected this particular scene toward a bluish tint to suggest coldness.  This is an artistic application of lighting.

Since this is a GOUT based project, I'd chalk the grainy skies up to a GOUT defect, and live with it along with the blinking stormtroopers.

Best to use just one setting for grain removal, and do not base the setting on the extreme sky scenes.   Live with grainy skies; optimize the grain removal for the rest of the film, avoiding plastic actors.

my humble opinion.

Doctor M said:

The other issue is the chapters points.  They are moved to the nearest I-frame.

Doesn't DVDLab Pro have a setting "Force I-Frame on Chapter Point"?

dark_jedi said:

Dufusyte said:

fwiw, the best practice for upscaling is to first upscale the raw data (preferably even larger than your final target), then perform all cleanup/tweak scripts on the overly large video, and then shrink the video down to your target size. 

The reasoning is:

1. Blow up the raw data (even "extra" blow it up); this preserves as much of the raw data as possible.  For example, if the raw data is 480, and your final target is 1080, then you might over blow it up from 480 to 2160 (twice the final target res) in this step.
   
   
2. Perform clean-up and tweaks on the blown up video; this performs the data improvement on the overly blown up video
   
   
3. Shrink back down to your target resolution (shrink from 2160 to 1080).  This final shrink will increase the sharpness and hide any artifacts that might have been produced by the tweaks in step 2.

 

The erroneous procedure would be:

1. First run the clean up and tweaks on the raw data.  This will reduce the raw data (lose some light and dark detail, etc)
   
   
2. Then blow up the tweaked video to the final target.  This will blow up any artifacts that were intoduced by the clean up and tweaks in Step 1.  Plus it will be blowing up degraded data, since some of the raw data was destroyed in Step 1.

This all sounds good but most of us don't really know what to mess with on g-force's script to do this, he said everything on it works together with each other the way it is, I just looked it over and really don't know where to even start to add those kind of settings, but I would definitely try it out if I did though, so right now I am doing everything afterwords.

I don't know the inner workings of the script either, and whether it is hardcoded for a specific resolution.  Sounds like the following suggestion would not be workable, but I will throw it out there anyway, since suggestions are being solicited.

1. Step 1: Upscale the GOUT using your favorite upscaler.
   
   
2. Step 2: Run the magic avisynth script to clean it up and tweak it.  (This might not work on an upscaled video if the resolution is hardcoded in the script though, or if certain parameters are resolution dependent)
   
   
3. Step 3: If the video was extra-upscaled in Step 1, then user your favorite down-scaler to downscale the video to the desired target resolution.
   
   

I guess my main suggestion is just to try running the cleanup script after the upres, rather than vice versa.  Hope this helps.

LeeThorogood said:

is that still just the raw GOUT upscaled or have you now moved onto upscaling the V3?

fwiw, the best practice for upscaling is to first upscale the raw data (preferably even larger than your final target), then perform all cleanup/tweak scripts on the overly large video, and then shrink the video down to your target size. 

The reasoning is:

1. Blow up the raw data (even "extra" blow it up); this preserves as much of the raw data as possible.  For example, if the raw data is 480, and your final target is 1080, then you might over blow it up from 480 to 2160 (twice the final target res) in this step.
   
   
2. Perform clean-up and tweaks on the blown up video; this performs the data improvement on the overly blown up video
   
   
3. Shrink back down to your target resolution (shrink from 2160 to 1080).  This final shrink will increase the sharpness and hide any artifacts that might have been produced by the tweaks in step 2.

The erroneous procedure would be:

1. First run the clean up and tweaks on the raw data.  This will reduce the raw data (lose some light and dark detail, etc)
   
   
2. Then blow up the tweaked video to the final target.  This will blow up any artifacts that were intoduced by the clean up and tweaks in Step 1.  Plus it will be blowing up degraded data, since some of the raw data was destroyed in Step 1.

I'd like to contribute some blank Verbatim DL disks to this worthy project.

msycamore said:

I would not call my subs safe from TV-overscan.

For commercial releases, the "Action Safe" zone is 10% inside the video border, and the "Title Safe" zone is 20% inside the video border.  Since flatscreens have less overscan than crt's, HD productions can use a smaller margin of error, say 5% for Action Safe and 10% for Title Safe.

Since there is a possibility a customer will play an anamorphic dvd on a 4:3 tv and choose to crop, the "Title Safe" zone would be, to ensure proper display in this worst case scenario, 10% inside the 4:3 portion of the video.

All of this is to guarantee, for commercial releases, that any words displayed in the video will be readable for all customers.

Since this is not a commercial release, feel free to do whatever you think reasonable.  I'd say you can assume the viewer is using a 16:9 screen (ergo, allow your subtitles to fail for the 4:3 crop scenario).  But I think it would still be good to observe the 10% parameter for Title Safe area inside the 16:9 frame, since this is kind of a standard, and a reasonable one at that.

The fact is that overscan is different on each device.

For your enjoyment, here is a graphic with the more conservative 4:3 safe area inside a 16:9 field displayed:
http://i206.photobucket.com/albums/bb41/joemorelli99/1080_sqpix_4x3_ttlsf.jpg

Here's a graphic with the 16:9 safe areas for 16:9 (would fail for a 4:3 scenario)
http://www.dvinfo.net/forum/final-cut-suite/144686-4-3-title-safe-16-9-comp.html

Here's a little discussion of the issues and a graphic in AfterEffects
http://toolfarm.com/reviews/aecs4.html

Here is more than anyone should ever know about safe zones:
http://provideocoalition.com/index.php/cmg_blogs/story/unsafe_areas/P1/

Televisions (especially Sony's) tend to push red anyway.  So be careful about too much red, since most tv's will further push the reds.

• To me 1134 looks kind of rubbery, as though the actors were plastic action figure dolls.  The skin looks rubbery plastic.
• 1159 is nice, but if you look at Han's black vest, you notice that the blacks are a bit crushed on Han's right lung side (our left, i.e., the side nearer to Chewy).  You can see more detail in the dark vest in the As Is.
• As Is looks a little bit purply and a little flat due to the low contrast.  But bumping up the contrast will crush some of the blacks, as seen above. 

I'd recommend using a soft touch with the tweaks.  1159 is near perfect, even with the slight loss of detail in the dark areas.  It's always a trade-off between the improved depth you get with High Contrast, vs the improved detail you get with Low Contrast.

I'd vote for either 1159 or As Is.  Or maybe halfway in between, i.e., dial back the 1159 tweaks a bit.  But at this point you can't go wrong with either.

Nowadays most tv's are LCD's and imho LCD tv's have exaggerated color representation anyway.  So even if the DVD is tweaked perfectly, the LCD TV will turn it into cartoon colors.  And most tv's come with default settings that are very aggressive, making the image "pop" with over saturated colors and high contrast.  This sells tv's in the store, but it's not very accurate color representation. Don't sweat the final tweaks, because they will be lost on most tv's.   Err on the gentler side, because most tv's will add Contrast and Saturation, especially reds. 

Great work and fine tweaks.  Ship it!

Actors have stage makeup applied to their entire face and possibly neck.  Meanwhile hands and chests do not receive makeup.  When looking at skintones, bear in mind that the face will always look a bit unnatural because it is all covered in foundation makeup (yes, even on the guys).  To see real skintones you would have to look at the hands or on Solo's chest, for example.  Anyway, that's why faces always look a little tan or odd - because of the stage makeup.

fwiw, film, like painting, is not necessarily realistic.  A director of photography will purposely cast different shots in different hues in order to convey different moods: bluish for coldheartedness, red for passion, yellow for warmth and joy, green for illness, etc. 

So one scene might be a different color cast than another for good reason, because of the art direction.

It would be erroneous to recolor each scene towards realism.  Realism might not be the intent of the film maker.

Having said this, I do opine that DJ's screenshots look good, and they appear to restore color which has been lost due to technical issues, not artistic issues.

The lettering "Original Theatrical Edition" at the top of these has jaggies.  Suggest using anti-alias on this lettering to avoid the jaggies.

In all other respects: a masterpiece.