Attract-Mode Support Forum
Attract-Mode Support => Scripting => Topic started by: Oomek on June 01, 2017, 06:09:56 AM
-
Attract Mode Blur Shader WIP - Snaps
https://youtu.be/FVazVqXL_bY
Attract Mode Blur Shader WIP - Videos
https://youtu.be/7x7etZvlvVs
-
cool,,cool,when can we start to use them?
were to get..
-
Not sure as I started it this morning, need to wrap it up in a module. Will take few days If I won't encounter any complications.
-
Looks great, awesome job! So you were able to do this strictly through shaders and a custom class/function in squirrel? I was under the impression that the blend modes couldn't be done without changes to the backend.. glad to see i was wrong :P
-
Blending modes could have been faster when used internally through the output merger, but limited to only 3 modes. I'm glad I've decided to go with shaders.
For the last few weeks I was busy building my first cab. Now when it's done I can come back to shaders and finish it all. Still not quite sure what algo I'm going to use for the blur, Gaussian or Kawase. Need some more testing.
-
Ohhh nice... :D
Blur is one of those things that can make text pop, which is important for low res monitors.
-
I don't follow. How can blur pop text out? You mean glow?
-
I don't follow. How can blur pop text out? You mean glow?
Anything close to a solid color makes text more readable. It's design 101. Take an image that is blurry and add text. Your eyes focus on the text instead of the image. The text is more defined at low resolutions where too much going on makes text illedgable.
-
Any update on this shader?
-
I can share if you are brave enough to clean it up and wrap into a class ;)
-
I can share if you are brave enough to clean it up and wrap into a class ;)
You got it! This week I revised the shader module. Hopefully it will be more useful for everyone! Will release before wednesday, my first day back to work.
-
It's a multipass kawase blur. It has a variable radius and deband which still needs to be tied to the radius. Atm these 2 parameters are independently set.
-
And obviously those 2 need to somehow scale to the level of pre-downsampling. I have't got time to figure out that part yet.
-
And obviously those 2 need to somehow scale to the level of pre-downsampling. I have't got time to figure out that part yet.
How do you implement pre-downsampling? I use the trick of placing the image on a surface of a desired size (so in the end for a two-pass downsampled filter I need two surfaces), is there a way you can do the downsampling in the shader?
-
That's exactly what I did. I draw a mimpapped image on a texture with reduced size and then use it as an input in the shader.