View Profile
View Forum Posts
Private Message
I've looked on Wiki, and they say 2K and 4K cinema are a relatively recent invention. It has been mooted that there was some sort of 'high def' copy made, in 1999, when the DVD's were released. Is this likely? I think they just took a standard DVD copy and upscaled it, as the standard and bluray are the same colour. I think this is going a bit further than Jagabo is saying,some 'intermediate standard', which no-one seems to have defined or seen working. Or, even, 'downscaled' bluray. It wouldn't surprise me if a lot of these Bluray releases are just processed standard DVD?
+ Reply to Thread
Results 31 to 50 of 50
-
-
You've stated that a number of times, and people have given their response to that. Post some links where you think the colour of the DVD and Bluray releases are the same. All of the images I've seen have clear differences in the colour reproduction.I think they just took a standard DVD copy and upscaled it, as the standard and bluray are the same colour.
Why do you hold the view that the films must be upscaled from standard def? Are you hoping that in the future the films will be re-scanned, and the result would be significantly better?
Unfortunatly, I'm not surprised that films of that age, which have been restored in a rather heavy handed way - excessive noise reduction, edge enhancement, etc look the way they do.
I'm not sure why the noise reduction is so high - I would've thought a modern h264/avc encoder at 35 to 40Mbps could easily render the full grain structure. It might be more to do with public expectation of a clean noise free HD image that has led the restoration team to do this.
What I would hope for, is a new Bluray release that has had minimal processing or manipulation. The Bluray equivelent of a Superbit DVD:
http://en.wikipedia.org/wiki/Superbit -
I am saying I think the films are just upscaled from standard DVD because:
1. It's cheaper. Going back to film is very expensive.
2. 2K and 4K cinema are recent inventions. Jagabo talks of some intermediate standard that may have existed when the films were made, or when the DVD's were copied. I have never heard of this.
3. On my Apple monitor the standard DVD and Bluray are roughly the same colour.
http://movies.trekcore.com/gallery/albums/tvhhd/tvhhd0241.jpg
http://movies.trekcore.com/gallery/albums/tvh/ch4/tvh0134.jpg
4. Jagabo mooted that they might 'downscale', to keep the public buying bluray, then release a full version. Why muddy the picture up, when you can just get programs that upscale from standard? I have seen them advertised on the net.
http://www.arcsoft.com/en-us/software_title.asp?ProductCode=SIMHD -
You're just trolling or joking, aren't you?Originally Posted by StewieTeabag
Cheers,
David. -
No, I'm not trolling or joking. They look similar enough to me, to raise suspicions. Every time I go back to my own photos, to rescan them, they come out slightly different, more different than these do.
http://www.arcsoft.com/en-us/softwar...ductCode=SIMHD
This costs about a hundred dollars. How much does the average film restoration unit have? -
You've gone from 'the same color' to 'roughly the same colour', and now 'similar enough'.Originally Posted by StewieTeabag
It might cost $100, but it can't convert SD to HD.http://www.arcsoft.com/en-us/softwar...ductCode=SIMHD
This costs about a hundred dollars. How much does the average film restoration unit have?
The motorbike example on the 'Features' tab looks fake - going from a few blurred pixels to readable 'SHOEI' lettering on the helmet, hmmm.
The YouTube example is so full of compression there's no way to judge what their software is doing. I used to use some of Arcsoft's software, it's a shame they're selling stuff like this now (or at least advertising it in this way). -
Actually, it only costs $10. But save your money and use MPCHC. It has similar runtime filters and it's free.
The Motorcycle image was faked. They produced the low res image from the high res image, not the other way around. The Butterfly image was sharpened, had the saturation increased, and the hue slightly shifted. Of course, they'll say these are just "simulations" for their web site.
The Youtube video is comparing the blurriest scaling algorithm (bilinear) to their sharper scaling and sharpening. As noted you can get similar results for free with MPCHC's Shader filters. Your graphics card's driver probably has similar filters.
Regarding the two Star Trek images: The high res image was not produced by upscaling a DVD source. The color differences are pretty large. But, in any case, if the DVD had accurate colors there is no reason the Blu-ray should have substantially different colors.Last edited by jagabo; 23rd Nov 2010 at 12:25.
-
Do you think it is possible that there are some processing programs that are not available to the public?
Where does NASA buy theirs from? Do they do it inhouse? -
Of course there are some processing programs that are not available to the public. But you can't bypass simple math.
I have two numbers. I'm going to compress them down to one number by averaging the two together to save space. That number is 57. What were my two original numbers? There's no way for you to know. Ever.
That is the fundamental problem with upscaling. The details that are not present in a small image cannot be perfectly restored when enlarging it. Sure, video has more information than my single number example. You can look at nearby pixels or pixels from previous and later frames to get some hints about what the upscaled image should look like. But you can never know for sure. The general algorithms used in upscaling enlarge the image and sharpen it, attempting to create as little artifacting as possible.
Look at sharp, nearly horizontal or nearly vertical edges. Sharp upscaling usually ends up displaying significant aliasing artifacts and over-sharpening halos. Cheap sharpening ends up increasing noise. Better sharpeners, usually called "edge enhancers", attempt not to sharpen noise, just edges.
Download the 720p version of this video from Arcsoft/Nvidia:
http://www.youtube.com/watch?v=egrC2WZbXQs
Look closely at the text. The upscaled version is full of oversharpening artifacts and enhanced DCT ringing noise:
Last edited by jagabo; 23rd Nov 2010 at 21:35.
-
Most of the 'enhancement' effect on that video can be achieved with standard tools. I often run mplayer with an 'unsharp mask' filter which does much the same thing.Originally Posted by jagabo
It looks like they've gone out of their way to find a film with subdued (but not non-existent) high frequency detail - and then boosted it.
mplayer before and after:
-
They're not similar at all to my eyes. YMMV.Originally Posted by StewieTeabag
On the same scanner? If so, I think your scanner is broken. If you're using different consumer scanners, then it's more likely, but at least one of them is possibly not very good! Remember: studio equipment is calibrated. Any differences (other than due to film ageing) could be intentional. A calibrated flat transfer would look almost the same, time after time. In practice, it's always tweaked on the way to disc, so different releases from the same master scan could look very different if someone decided to colour-correct, denoise, sharpen etc differently each time.Every time I go back to my own photos, to rescan them, they come out slightly different, more different than these do.
As others have rightly said, you can't re-create genuine details from nothing. You can take weak details and boost them, but the BluRay vs DVD comparisons you've linked to are far more than that.http://www.arcsoft.com/en-us/softwar...ductCode=SIMHD
This costs about a hundred dollars. How much does the average film restoration unit have?
You might find this interesting - it's a comparison of various commercial "super resolution" techniques. I put that in quotes because they're not really, but they do upscale nicely.
http://forum.doom9.org/showthread.php?t=154141
If you want to see more options, check out these threads...
http://forum.doom9.org/showthread.php?t=152623
http://forum.doom9.org/showthread.php?t=145314
http://forum.doom9.org/showthread.php?t=142704
http://forum.doom9.org/showthread.php?t=109422
Some fascinating stuff, and some great results. Far better than a dumb resize, but still clearly worse a native HD source.
You can also google Super Resolution, e.g. http://en.wikipedia.org/wiki/Super-resolution but understand that the truly miraculous super resolution results only happen when the source has lots of aliasing, and the underlying image or scene doesn't change (i.e. the camera just pans across unchanging scenery). Neither is true of a typical DVD.
Now think carefully about this: You came here, asked a question, got given the same answer by several different people (with completely different backgrounds, sitting in different countries!), and now you want to argue with that answer.
Why did you ask the question if you'd already decided what the answer was? If it's to learn and understand, that's fine - but please realise that if people give you an answer and you simply don't believe them, then they may feel like they're wasting their time talking with you. That would be a shame, because you've received some great replies in this thread.
Maybe it should be renamed "SD>HD upscaling / super resolution" and made a sticky!
Cheers,
David. -
I have received some great informative replies.
Thanks. -
What can be useful is to start with a clean, sharp, detailed 1920x1080 HD source, reduce it to 720x480 DVD resolution, then enlarge that back to HD size using different upscalers. Compare the final result to the original HD source. You'll see that even the best upscalers are nowhere near the original HD source.
Intracube showed a few examples using a primitive upscaling algorithm earlier:
https://forum.videohelp.com/threads/328291-re-A-little-balance-please?p=2033278&viewful...=1#post2033278 -
Are we talking about a specific movie or in general for DVD or Blu-Ray commercial releases?Originally Posted by StewieTeabag
You seem to not understand the film transfer process. Let's try again.
Flying spot film scanners (e.g. Rank Cintel) have improved in resolution over the years. The late 80s - early 90s top end units output to Rec.601 4:2:2. These were color corrected directly off the scanner before recording to D1 or later DigiBeta tape. These were the 'standard definition" masters.
Early high definition transfers were color corrected (e.g. DaVinci) then either recorded to a digital intermediate or 1440x1080 HDCAM tape (introduced 1997). Target distribution for these transfers was HDTV broadcast, early digital cinema, future HD consumer optical disc (Blu-Ray introduced 2003) and for downscale for DVD release.
Between 2003 and 2007 the standards for D-Cinema, Blu-Ray and HD DVD were finalized. The common transfer and mastering standard became 2K or 4k digital intermediates for major films. TV series used 1920x1080 transfers.
While it is possible to make a Blu-Ray disc from D1 or HDCAM masters, this was not typical for new film releases. They were freshly transferred to 2k, 4k or proprietary digital intermediates. Most earlier films were freshly transferred and remastered for Blu-Ray release. If any of the Star Trek movie Blu-Ray releases were SD upscales we would know about it and could detect it.Recommends: Kiva.org - Loans that change lives.
http://www.kiva.org/about -
They don't look similar to me.Originally Posted by StewieTeabag
Your "DVD" image isn't a representative frame from the DVD VOB, it is a 850x358 upscaled processed image done either in you software player or display card.
Your "Blu-Ray" frame has also been processed.
Why not obtain frames directly from the media for comparison?Recommends: Kiva.org - Loans that change lives.
http://www.kiva.org/about -
More important for upscale is access to the original uncompressed 270 Mb/s 10 bit SD digital master, not the ~4Mb/s 8 bit MPeg2 compression released to DVD.Originally Posted by StewieTeabag
There are many proprietary methods for upscale.Recommends: Kiva.org - Loans that change lives.
http://www.kiva.org/about -
Those screencaps Intracube posted in message 40, seem to me to give some indication of what they've done to the ST films. Just brought out a few more pores and wrinkles with a bit of processing. You can really tell the difference, when you look at ST2, and the rest.
-
No, they don't, because those are at the same resolution. If you start with a DVD image, upscale it, and then try to sharpen it, you will have blocky mush. If you sharpen then upscale, you will have large halos.Originally Posted by StewieTeabag
Cheers,
David. -
Thanks for the interesting links.Originally Posted by 2Bdecided
Most of the up-scaling examples exhibit aliasing and or halos. Although there are a few examples that look noticeably cleaner and free from artifacts.
The example in the last link (of an artist painting a castle) is clever, but as you say, it exploits an un-anti aliased image. Most pictures would be filtered to at least half the sampling frequency (Nyquist). I'd also expect to see funny artifacts if this was tried on a video or a more complex scene.
The castle (Neuschwanstein Castle) is interesting:
http://forum.doom9.org/showthread.php?t=152623
I'm still sceptical about that one. Take a look at the middle window:
I can't see any evidence of the central pillar on the 'original' image (1). Yet the up-scaled image shows a very clear and defined pillar (2).
The second example shows little to no detail on the tops of the columns (3) - particularly within the white boxes, yet the up-scaled image (4) shows 3 discrete tops to the columns:
The bottom image was taken from Wikipedia:
http://en.wikipedia.org/wiki/Neuschwanstein_CastleLast edited by intracube; 25th Nov 2010 at 11:47.
-
Note that motion film or video upscale has additional issues vs. stills. Frame to frame issues caused by noise or motion estimation errors result in flickery artifacts that wouldn't be noticed in stills.
Recommends: Kiva.org - Loans that change lives.
http://www.kiva.org/about
Quote
Similar Threads
-
Please share your White Balance methods.
By thurnau in forum Newbie / General discussionsReplies: 10Last Post: 28th Nov 2010, 02:59 -
white balance-sample
By kzeko in forum RestorationReplies: 20Last Post: 22nd Oct 2009, 19:06 -
Need to balance audio
By grayprint in forum RestorationReplies: 2Last Post: 25th Feb 2009, 13:51 -
Finding the balance (x264)
By doggyjumper in forum Video ConversionReplies: 27Last Post: 8th Dec 2008, 12:04 -
How to balance the sound level?
By coody in forum AudioReplies: 2Last Post: 30th Jun 2008, 21:18