I have to remove aliasing from this anime, is it possible?

[Myfigure]

Thank you!

I have this with supaa():



There is less aliasing, right?

[Myfigure]

Anyway, I see banding. Example:



I have to use debanding?

[Scintilla]

Anyway, I see banding. Example:

Quoted Image converted to link:
http://imageshack.us/photo/my-images/807/banding.png/

I have to use debanding?

It says the image is supposed to be 1280x720, but I can't seem to view it at that size -- I click the "Zoom" button on Imageshack, and the page that loads has the image at exactly the same size (just with 100% more annoying popup advertisements).
At the smaller size, I can't see the banding problem.

[Myfigure]

Anyway, I see banding. Example:

Quoted Image converted to link:
http://imageshack.us/photo/my-images/807/banding.png/

I have to use debanding?

It says the image is supposed to be 1280x720, but I can't seem to view it at that size -- I click the "Zoom" button on Imageshack, and the page that loads has the image at exactly the same size (just with 100% more annoying popup advertisements).
At the smaller size, I can't see the banding problem.

Try this: http://imageshack.us/m/805/4743/banding2.png

[Cannonaire]

I don't see banding, but I do see blocking, which is most likely a result of the compression. It looks fairly minor in the screenshots, but as with many video artifacts it might need to be seen in motion to truly gauge how much of a problem it is. If it is an issue, you can try a deblocker on a low setting. I've only used deblock(), but it has worked well for me so far. Perhaps someone else has another suggestion or reasons why deblock() might not be a good idea.

Because the footage seems to be prone to blocking, you might want to read up on the --deblock flag in x264 ('deblocking' settings in Zarx264gui) when it comes time to encode your video for release. If you do it right and you also aren't too aggressive with quantization, you can keep the blocking from appearing again in the final product.

[Myfigure]

Thanks for your answer.

With these screens I used --deblock -1:-2 and CRF 15.

[Mister Hatt]

I see no banding either and not much blocking, although I'm on a 6bit LCD right now so it could have been dithered over. You did mostly kill the 'aliasing' at least, but seems you ignored the rest of what I said aside from using supaa(). Bicubic or even bilinearly downscaling and then pp'ing it with EE() would look a lot better, and probably fix any banding or blocking you might experience unless it's caused by quantisation.

[mirkosp]

I'd say adding a very slight grain over the thing would likely resolve the blocks you're seeing there. I think it's not in the source but a quantization issue with the encode. Possibly increasing the aq-strength could help as well or actually fix on its own.

[Myfigure]

What is EE? I don't see this filter on avisynth wiki. Anyway, I don't have unsharp mask for this filter.

Here with aq-strength 1.0 (I had 0.6). I think that now I have fixed these blocks.

http://img508.imageshack.us/img508/4525/aqstrength1.png

Here screen from BD, is there a little grain right?

http://img861.imageshack.us/img861/6282/bd1.png

[Mister Hatt]

0.6 is far too low. Depending on what causes the blocks, you either want to go 0.8-0.9 or possibly 1.2, but it depends on the cause. As I said earlier though, downscaling the correct way will probably fix the cause of the blocks anyway (and I am far too tired to explain why).