This screenshot is interesting, because piota recently posted a UHD scan of the same frame.
Avisynth Spline64Resize vs SRV11 raw
http://screenshotcomparison.com/comparison/130459
SRV11 raw vs 35 mm
Could you resize that comparison please. :)
I second seeing these with a dehalo added
Preferred Saga:
1,2: Numeraljoker extended
3: L8wrtr
4,6-9: Hal9000
5: Adywan
I just quickly downloaded and resized them with PS. Resizing might affect the quality, but there is no further compression as I saved in PNG.
Darth Id on ‘Why “Ben”?’:
And while we’re at it, we need to figure out why they kept calling Mark Hamill’s character “Luke Skywalker,” since it’s my subjective opinion that his name is actually Schnarzle Shnuzzle. It just doesn’t make sense!
Damn you George Lucas for never explaining why they all keep calling Schnarzle “Luke”!
Damn You!!!
Impressive. Both look nice in their own way. If you got rid of the halos they would be closer. It almost seems like there is more detail in the round dark glass part of R2 in the SR version. There is less grain in the SR version too. The aspect ratio is slightly different when looking at the square / tall and skinny grill in the front.
DrDre said:
Here are a couple of screenshots at 1080p, without degraining, dehaloing etc.
http://screenshotcomparison.com/comparison/130449
http://screenshotcomparison.com/comparison/130450
OMFG
If your crop is water, what, exactly, would you dust your crops with?
Here the comparison to the scaled 35 mm frame, with all the post-processing, but without degraining.
SRV11 raw vs SRV11
http://screenshotcomparison.com/comparison/130486
SRV11 vs 35 mm
http://screenshotcomparison.com/comparison/130487
The AR is still a matter for debate. The current AR closely matches the AR used by Team Blu, aside from some cropping (probably due to stabilization).
There is a downside to the dehaloing though, that is pretty obvious at 1080p. The detail and clarity are clearly affected:
Some of the other frames you posted had more obvious haloing. Can you share more comparisons using the same SRv11 raw frames you posted earlier? My initial impression is it is not worth it using this current method, but it is hard to say at this point.
Halos, a type of echo of portions of the image, within the image. Mostly prevalent at edges with high contrast.
Suppose there is a way to leverage SR technology to combat halos? If SR normally works by finding similarities in images and combining detail to strengthen true image information, could it not also be run "in reverse" to detect lower-detail echoes and suppress them instead of enhancing them, while still improving in true detail?
If your crop is water, what, exactly, would you dust your crops with?
I think I found a setting that affects most of the haloing, but leaves ~99% of the detail intact. Here are a few 720p screenshots:
http://screenshotcomparison.com/comparison/130571
http://screenshotcomparison.com/comparison/130572
Thanks for the screenshots. In my opinion, it isn't doing very much to eliminate the halos, but it is detrimentally affecting the image. The dehaloed are more dull, less shiny, less detailed and sort of look like looking through dirty glass. The raw images are better. I can live with the halos much more than the aliasing so it is not worth ruining the crispness and 3D pop to the picture to eliminate the halos. Maybe there are other methods that work better. If it could be combined with the edge detection, that might help. The halos tend to happen at the edges of the objects.
Not sure if this helps or not: http://forum.videohelp.com/threads/371768-Dehalo-and-Chroma-Sharpening?p=2389569
I have to agree with thorr, the de-haloed images just don't remove enough of the haloing to counteract the loss of detail in many areas. It does tend to eat into the 3D image.
Although it does seem worth it to dive into the de-haloing process more and improve upon it! :)
The halos are reduced, but it indeed affects the detail and depth as well. I have been using the function that's mentioned in the thread. It works on the edges, but defining an edge is not an exact science.
Here are a few more 1080p screenshots without dehaloing:
http://screenshotcomparison.com/comparison/130594
http://screenshotcomparison.com/comparison/130595
http://screenshotcomparison.com/comparison/130596
http://screenshotcomparison.com/comparison/130598
http://screenshotcomparison.com/comparison/130599
Looks really great! Can't wait to see it in motion!
Regarding the aspect ratio, I thought of an easy way to check it. Look at the grill on R2D2 and google images of R2D2. It is supposed to be a bit taller than wide similar to poita's sample frame. So it looks like your version should be stretched vertically a bit.
Here is a comparison with R2 in nearly the same orientation with the one in the middle stretched vertically 106%: http://i57.tinypic.com/2iuouip.png
Those latest images are really excellent. One thing I noticed though, particularly in the final two images, is that objects in motion tend to lose their detail, such as Luke's hair as it is moved by the wind, and some detail on the speeder in the last comparison.
You probably don’t recognize me because of the red arm.
Episode 9 Rewrite, The Starlight Project (Released!) and ANH Technicolor Project (Released!)
It could just be motion blur in the frame. Seeing the HD version for comparison would make sure of it.
Regarding the aspect ratio, I checked the distances of the pixels in my comparison and came up with 106.17% vertical stretch. The 106% that I eyeballed was close, but still barely too short. So instead of 1920x816, it should be 1920x866 then add black bars above it and below it to make it 1920x1080.
I hope I'm not being too negative or outdated here in saying...
I really like the compromises of aliasing and detail in v10. :)
Actually the aliasing has been significantly reduced compared to v10, and v11 has a lot more detail. There is more ringing and grain, as this seems to be preferred above less detail and depth.
Here are the two frames mentioned compared to Harmy's DE. It is true that the antialiasing affects the detail negatively in places, but I think it is relatively minor compared to the detail corrected by the antialiasing.
DrDre said:
Here are the two frames mentioned compared to Harmy's DE. It is true that the antialiasing affects the detail negatively in places, but I think it is relatively minor compared to the detail corrected by the antialiasing.
Socks knocked off.
In addition to the SR magic...
The GOUT image... It's like the original telecine process was transfered at 16:9 anamorphic and was converted to 4:3 letterbox at the mastering stage (and caused all of that awful aliasing) by throwing out every other scanline like yesterday's bagels.
I don't know how, but I believe you found all those scanlines that they tossed away, then rolled them in sesame seeds, added the most amazing cream cheese, homestyle berry preserves, smoky lox and toasted it all so that is 9 times better than when the bagels were fresh to begin with. That's how good this work is.
Truth is, I don't even know if 16:9 anamorphic standard definition video was even used during the (pre-DVD) Laserdisc era for feature film transfers, so I don't believe that those "missing scanlines" of the GOUT were ever even in any video signal from the telecine machine on down the video chain. I know HD video had been in development for several years at the time of those film transfers, but pretty much as R&D only.
Still, in the back of my mind, I wonder if there might be a 16:9 anamorphic D-1 tape straight from the Star Wars telecine session floating around. If a very early scan converter was used to convert this (hypothetical)16:9 anamorphic film transfer to 4:3 letterbox format, and that subsequent 4:3 tape was used as the Laserdisc master, that could explain the heavy aliasing of the GOUT. If Lucasfilm returned to these scan-converted transfer tapes to produce the 2006 bonus DVDs (erroneously believing them to be the best existing transfers of the Original Trilogy), they had missed the opportunity to use these hypothetical anamorphic original transfer tapes instead.
If your crop is water, what, exactly, would you dust your crops with?
Thanks for the compliments! I'm no expert on Laserdisc transfers, I think Laserdisc Master, Althor1138, Anctufaalb or a number of other experts could probably shed some light on the subject. I'm just glad that the information about the high resolution frames is still hidden in the low resolution frames. As one of the members noted a while ago, it's precisely the large amount of aliasing that probably makes it work so well.
I will do some more tests with respect to the AR. Although the R2D2 frames suggest they need to be stretched vertically, some of the frames with C3PO seem to suggest it needs to be stretched horizontally when compared to Harmy's DE.
camroncamera said:
The GOUT image... It's like the original telecine process was transfered at 16:9 anamorphic and was converted to 4:3 letterbox at the mastering stage (and caused all of that awful aliasing) by throwing out every other scanline like yesterday's bagels.I don't know how, but I believe you found all those scanlines that they tossed away, then rolled them in sesame seeds, added the most amazing cream cheese, homestyle berry preserves, smoky lox and toasted it all so that is 9 times better than when the bagels were fresh to begin with. That's how good this work is.
Truth is, I don't even know if 16:9 anamorphic standard definition video was even used during the (pre-DVD) Laserdisc era for feature film transfers, so I don't believe that those "missing scanlines" of the GOUT were ever even in any video signal from the telecine machine on down the video chain. I know HD video had been in development for several years at the time of those film transfers, but pretty much as R&D only.
Still, in the back of my mind, I wonder if there might be a 16:9 anamorphic D-1 tape straight from the Star Wars telecine session floating around. If a very early scan converter was used to convert this (hypothetical)16:9 anamorphic film transfer to 4:3 letterbox format, and that subsequent 4:3 tape was used as the Laserdisc master, that could explain the heavy aliasing of the GOUT. If Lucasfilm returned to these scan-converted transfer tapes to produce the 2006 bonus DVDs (erroneously believing them to be the best existing transfers of the Original Trilogy), they had missed the opportunity to use these hypothetical anamorphic original transfer tapes instead.
No, the aliasing is just a side-effect of the telecine process. One field is scanned, then the other, the data is written straight to videotape and the process is repeated until the film is done. If the odd and even lines are slightly misaligned then you get aliasing.
[ Scanning stuff since 2015 ]