Image processing on either GD?

[Mikelemec]

I apologise in advance for my ignorance.

From what I understand sprites / bitmaps only interact with themselves in terms of collisions / blending and not with the background. I wondered whether it would be possible to apply techniques used for "old" VGA type games for image processing. For example, would it be possible to modify small sprites or bitmaps on GD1 or 2 to interact with the background or a background sprite. Like adapting the sprite to be a tiny convolutional filter. I am not looking for conventional use of the filter either. I guess it would be moving a GPU to towards a GPGPU. Does this make any sense at all or just defeat the whole point? I'm guessing this will kill performance.

[jamesbowman]

Yes, it is possible - in quite a limited sense!

For example the stencil buffer can be used to do 8-bit computations. Something like this is definitely possible:

what-when-how.com/opengl-programming-guide/life-in-the-stencil-buffer-opengl-programming/

The only expensive operation is bitmap readback. The "write" passes would be really fast, about 1 billion operations per second on GD3.

[Mikelemec]

Thanks for the link to Game of Life. I found another example using a stencil to perform a Gaussian blur on an image. This is all new to me so I need to get up to speed on all things EVE and get comfortable with what's possible.



Searching around I found mostly things "GD2". Am I right in thinking the "GD3" is the library associated with the Teensy board and the quoted speed involves using it?



Depending on the response to my last question, I would appreciate advice on hardware to try out some ideas. Based on looking at posts in the community it seems that a"Redboard" arduino and an FTDI dev board such as FT810 with 5" display should be ok.. Should I consider anything else?



Sorry for all the questions!