Originally published at: Enjoy this upscaled 'Thundercats' intro | Boing Boing
…
My computer did not like trying to run it at 8k, but at 4k it still looked great! Might try later with my personal computer.
Thundercats was a great cartoon. Though I am still scarred/confused with the flash back episode where they were all basically naked??
That one is my fav, so honest, so real, it took courage.
8k…sometimes I feel like people really don’t understand how video formats and resolution work.
Not saying 1080 is good enough by any means, but very few people have the hardware and viewing distance to tell the difference between a good 1080 source and 4k.
Is it just me or did Cheetara fuel some strange adolescent fantasies?
Looks OK, I guess. Some frames of weirdness going on, but at least it wasn’t one of those awful “60 fps” videos using AI to ruin animation. Ugh.
Don’t all youngsters daydream about bringing down gazelles?
Does anyone here even have an 8K monitor?
That opener never fails to give me chills for some reason - in a good way. Same way as the “Circle of Life” scene in TLK, Trans-Siberian Orchestra’s “Christmas Eve/Sarajevo,” and the t-rex’s triumphant roar in “Jurassic Park” does. No idea why.
Oh man that looks horrible. All the lines look like they were shoved though usharp mask.
The weird artifacting around lines are odd and the way film grain kinda shows up in places is just … weird.
This reminds me of the horrible shit they did to 101 Dalmatians and Sword in the Stone bluray transfers where they shoved it in an automated cleanup routine that destroyed all the line-work and gave this kind of effect.
I find it amazing the computer people that do things like that that don’t really understand the media they are screwing with.
Ick:
I didn’t even watch it until your comment.
(I mean I don’t even have a 4k monitor…)
Not like that matters, it still looks like hot garbage.
I think 2 problems exist. The first one is the source material. If you could get the physical animation film or cells and then obtain a +4k scan of them you’d have a really really solid basis to clean up the source from. More than likely someone is working from a broadcast tape of some sort (if they are lucky). Worst case they are taking some over compressed sd res video they found on the net and started there. Second, I think the vast majority of people have a very skewed idea of what is supposed to look good. It’s like going to Best Buy and viewing TV’s. All of them have the brightness, contrast, and saturation blown completely out. And then most of them are playing some type of “stock” promotional footage that is optimized to the TV. Back in the late 90’s I was at Circuit City trying to pick out a new aftermarket CD player for my car. I asked the sales guy to flat line a couple of them, zero out the bass, mid, and treble and turn off any of the enhancements. He looked at me like I grew a second head.
I agree with everything you said. It’s overly sharp, yet the edges and lines are soft and distorted. There is always going to be a big difference between filming something in +4k, taking an analog source and capturing it in +4k, and using a computer to upscale to +4k. Even 1080p today looks better than it did 20 years ago just because camera sensor technology has progressed that much. The end goal of these sort of projects shouldn’t be to give you a true 8k experience, cause that’s not possible. It should be to make a video that works well on the current tech platforms (which is 4k at best) and make it look like a better version over what a standard resize/scale would do.
Cause really, with a clean source I could probably do better with nothing more than Gigapixel and Avisynth. (Also I have no idea what the actual bitrate Youtube is throwing out for 8k.)
It’s true, the whole “Oh look! AI did this and made it 4K at 60fps” you get the impression of some sort of blade runner thing where it managed to fill in and figure out the gaps well. But the people doing this don’t understand the material. It should have film grain, hell if they had taken this output and shoved in it after affects and just did some slight simulated grain it might have sold the results better or at least made the output have a consistent look. Yeah it’s better than just a resize but … neither is actually that good.
I have the same beef with most deep fakes I’ve seen. The person started with really shitty overly processed video to being with and the results tend to be weird.
I wouldn’t want to yuck anyone’s yum, but the logic of “smart” upscaling really escapes me. If you display an image using 64 times as many pixels as it actually has, then you’re making up 98% of what the viewer sees. The most faithful way is to simply enlarge the real pixels; if you try to invent new pixels, you can only be adding a form of noise, and doing it in a “smarter” way just makes it uncannier.
As far as I can see it’s about making the display look better at the expense of the thing being displayed. So like, you see a sharper picture than your old TV could have displayed, but it has to make the picture look worse to prove that point.
Almost as sharp as I “remember” it as a kid…through my own Nostalgiamatic-16k filter.
It irks me that people keep labeling AI-upsampled videos as "remastered’. That’s not what that word means, there was no master copy involved, they just fed a standard-def YouTube video through a filter.
Er… I really don’t see an issue with the Tigra image. It is nice. And I didn’t notice it at the time, but the airbrushed background looks like you can touch it.