devolution

Author Topic: General Discussion  (Read 3821761 times)

0 Members and 23 Guests are viewing this topic.

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8085 on: August 14, 2016, 10:47:30 AM »
Actually i use YUVA and not RGBA as its the only way to encode it right with the solution iv found
« Last Edit: August 14, 2016, 10:52:43 AM by lamoli »

Offline DarkTl

  • Hero Member
  • *****
  • Posts: 4737
Re: General Discussion
« Reply #8086 on: August 14, 2016, 10:51:28 AM »
Here a test..
We usually use "pytfall" password for archives, but in your case it doesn't work, so I'm at loss here  :)

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8087 on: August 14, 2016, 10:52:21 AM »
We usually use "pytfall" password for archives, but in your case it doesn't work, so I'm at loss here  :)


Omg must have used "Pytfall" this time .. i just renamed hundreds of files with Capitals so it must have been a mistake from there lol
« Last Edit: August 14, 2016, 10:54:07 AM by lamoli »

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8088 on: August 14, 2016, 10:54:30 AM »
I think it works differently in After Effects. I set colors channels to alpha without rendering transparency. That results in black and white webm where pure white is alpha 255 and pure black is alpha 0. At this point doesn't matter which color channel is used by renpy for mask, since they all present in white color equally.

But if I disable some channels instead of including them into alpha, that might lead to not including parts of animation into the black and white webm, so renpy will not even see them.

Well, like I've said, you may be doing some custom propitiatory thing. ffmpeg can actually do a mask without showing anything visible at all... some weird internal webm thing. What I laid out there is how it is working logically, you have four channels in .png sequence and four channels between the two webm movies. in the second movie, blue and green channels are not read by Ren'Py so if whatever you're doing encodes red channel to be what it got to be, it's all perfectly valid.

Here a test..


https://www.dropbox.com/s/8sqjmygnnn9e95t/WEBM.rar?dl=0 Run the html with chrome ( webm doesn't follow resize but you can drag it to where it should be if you do change the window size )

As for the webm itself its not a animation its a morphing from pose to pose and i asked blender to extrapolate the transitions.. but know that i see it can do all this i might make some real animations..

Password?

As I've wrote like 3 times already, webm with it's own Alpha will play in Ren'Py just like a webm without it. You need a mas movie if you want it to handle alpha as well.
Like what we're doing?

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8089 on: August 14, 2016, 10:57:42 AM »
Its "Pytfall" mistakenly use caps P


i would realy love to get my hands on the chrome transparency class code :p
« Last Edit: August 14, 2016, 10:59:28 AM by lamoli »

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8090 on: August 14, 2016, 10:59:42 AM »
Its "Pytfall" mistakenly use caps P

Got it. Well, it works in Chrome and nowhere else. Even Firefox doesn't show it... there is not even a chance that we can show that in Ren'Py but we can make a masked movie from .png sequence or just used still .png frames and the background.
Like what we're doing?

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8091 on: August 14, 2016, 11:04:08 AM »
Got it. Well, it works in Chrome and nowhere else. Even Firefox doesn't show it... there is not even a chance that we can show that in Ren'Py but we can make a masked movie from .png sequence or just used still .png frames and the background.


indeed even VLC doent work as what you need to do is:


http://wiki.webmproject.org/alpha-channel and of course no body did that other than chrome




the raw frame counts tops at 1.2k (526mb) so no way while webm is a few mb ( give us animated gifs back as webm is useless atm lol !! )


Well not sure if you can mask 1.2k pngs >.<
« Last Edit: August 14, 2016, 11:09:36 AM by lamoli »

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8092 on: August 14, 2016, 11:08:18 AM »
i would realy love to get my hands on the chrome transparency class code :p

Like: https://chromium.googlesource.com/webm/webm-tools

Vignesh Venkatasubramanian is the guy in google who is responsible for continues support. Rest of the pushes is by the support team. I think that is the part related to webm alpha channels but there is (always) more relevant code.

I'd not even consider bothering with this crap:

- Figuring out their approach.
- Rewriting encoder that is used by Ren'Py.
- Rebuilding Ren'Py (it's difficult, like I had to spend half a day installing all that's needed and figuring out how to make just that work few years back) and using our own custom version after every major upgrade.

it's not even adjacent to a realm of being worth the effort while we have a perfectly working alternative!
Like what we're doing?

Offline DarkTl

  • Hero Member
  • *****
  • Posts: 4737
Re: General Discussion
« Reply #8093 on: August 14, 2016, 11:10:52 AM »
Formats war... Just like apng was developed and mostly supported by mozilla  :)

In this case we can't even use the image itself as a mask, like for some spells. It becomes too transparent.

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8094 on: August 14, 2016, 11:11:48 AM »
Yeah unless they fix it best would be masks.. but that needs 1 files right ?

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8095 on: August 14, 2016, 11:13:20 AM »
the raw frame counts tops at 1.2k (526mb) so no way while webm is a few mb ( give us animated gifs back as webm is useless atm lol !! )

BS...

I've proven couple of hundred posts back that WEBM resulted in MUCH better colors and smaller size than .gif with the best encoding... look at our god damned animations, WEBM beats both .png sequences and .gif in quality, size or both.

You just make two damned movies, one as a normal .webm and the second one with alpha channel encoded to red channel of the movie. Two movies so far ALWAYS resulted in smaller size and better quality than a single gif! It's just a better, more modern format... there is no way around that plain and simple fact!
Like what we're doing?

Offline lamoli

  • Hero Member
  • *****
  • Posts: 553
Re: General Discussion
« Reply #8096 on: August 14, 2016, 11:16:46 AM »
BS...

I've proven couple of hundred posts back that WEBM resulted in MUCH better colors and smaller size than .gif with the best encoding... look at our god damned animations, WEBM beats both .png sequences and .gif in quality, size or both.

You just make two damned movies, one as a normal .webm and the second one with alpha channel encoded to red channel of the movie. Two movies so far ALWAYS resulted in smaller size and better quality than a single gif! It's just a better, more modern format... there is no way around that plain and simple fact!


I do agree only problem now is format will change for sure so does renpy code and in the end it might be hard to make everything work all the time i guess.. ill use Dark 2 webm to make compatible one now..

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8097 on: August 14, 2016, 11:21:36 AM »
In this case we can't even use the image itself as a mask, like for some spells. It becomes too transparent.

Yes you could...

Formats war... Just like apng was developed and mostly supported by mozilla  :)

Actually apng is a really decent sh!t... PyTom think so too and he looked into it, Ren'Py can support it, it just requires time and effort to implement and he said that he'd do it only if it trully went mainstream. There is little point in it right now becausem even less people know how to encode apng than .webm :)

Yeah unless they fix it best would be masks.. but that needs 1 files right ?

Look at our animation and their sources. Quality of Ren'Py webm implementation is just as good as .png sequence. At times, it is not even possible to do a .png sequence that can be rendered at the same speed as .webm by Ren'Py (but we do not have frame-rates that are anywhere near that). Size so far has always been smaller (sometimes likes up to 8x smaller than .png and 50% - 400% smaller than .gif).

I've explained it really well above, you need one, plain and simple .webm and another .webm where alpha channel is encoded to red channel.
Like what we're doing?

Offline Xela

  • Global Moderator
  • *****
  • Posts: 6893
  • "It's like hunting cows"
Re: General Discussion
« Reply #8098 on: August 14, 2016, 11:30:15 AM »
I do agree only problem now is format will change for sure so does renpy code and in the end it might be hard to make everything work all the time i guess.. ill use Dark 2 webm to make compatible one now..

Ren'Py really goes out of the way to keep sh!t backwards compatible... to an unreasonable (imo) degree. So whatever works now will work for a long time. PNG sequence is the safest bet but I had one weird case (prolly poorly encoded .pngs) where the size went down 20 times, from 120MB to 6MB (both movies). There was no quality loss that I could notice.

We're not releasing for mobile and I don't know if .webm works in Ren'Py on iOS but they were not (officially) supported on Android, however we've tested it in a different game and it worked absolutely perfectly...
Like what we're doing?

Offline DarkTl

  • Hero Member
  • *****
  • Posts: 4737
Re: General Discussion
« Reply #8099 on: August 14, 2016, 12:03:42 PM »
Yes you could...
I tried it actually with the last lamoli example, the girl is half transparent in this case.