| |
polyp2000
Account closed
Registered: Dec 2016
Posts: 5 |
Using NUFLI / FLI Etc for animation
Anyone know what kind of scope there is for "modifying"
the contents of these pictures on the fly ?
How about something really simple like a changing a section to a solid block of color |
|
| |
JackAsser
Registered: Jun 2002
Posts: 2014 |
For FLI it's trivial. Just sta 8 color values to get a solid char |
| |
Oswald
Registered: Apr 2002
Posts: 5086 |
Quote: For FLI it's trivial. Just sta 8 color values to get a solid char
in case of nufli sprites will come in the way + you forgot d800 |
| |
Cruzer
Registered: Dec 2001
Posts: 1048 |
Animation as in prerendered stuff or realtime effects? |
| |
Bitbreaker
Registered: Oct 2002
Posts: 504 |
Also FLI content in NUFLI is changed in even and sprite stuff in odd lines. Have fun determining where the block you want to manipulate shall start/end, as you always interfere on the next line either with FLI or sprite. |
| |
Mr. SID
Registered: Jan 2003
Posts: 424 |
I guess you could try making two NUFLI images (one normal, the other one modified), and compare the data to see what differences are.
And then generate a small piece of code that will apply those changes. |
| |
Oswald
Registered: Apr 2002
Posts: 5086 |
Quote: in case of nufli sprites will come in the way + you forgot d800
okay no d800 in nufli O:-) |
| |
ChristopherJam
Registered: Aug 2004
Posts: 1408 |
Yeah, real time generated content's going to be hard; allocating colour attributes is non trivial, and for anything beyond replacing one FLI 8x1 at a time you're going to need to either track recent requests or mark pixels as "I'm going to replace this one soon so IDK if you trash it for a bit" |
| |
algorithm
Registered: May 2002
Posts: 705 |
Mr SIDS implementation would work well for NU-FLI. The method of rendering separate images and then marking the changes and then updating these per frame. Although ideally would be beneficial to use non dispersed dithering if possible. Would have around 7000 cycles or so left per frame to make the relevant changes |
| |
polyp2000
Account closed
Registered: Dec 2016
Posts: 5 |
If such modes are implemented on only a portion of the screen - does this free up sprites that can be animated? |
| |
ChristopherJam
Registered: Aug 2004
Posts: 1408 |
Quote: If such modes are implemented on only a portion of the screen - does this free up sprites that can be animated?
Yes. |
| |
Repose
Registered: Oct 2010
Posts: 225 |
This remotely helps, but I solved the algorithm for optimal colouring of FLI, it can be done very quickly. Has anyone else found this solution?
Now don't ask me exactly how it works right now, but I had to count colours in special way. It had something to do with unique colours per 4x8 block and subtract that from the most frequent color and then that's your colour nybble.
I never could find a good solution for IFLI though, however my algorithm was a decent start point, but not always optimal.
Btw there is always the problem of the errors (the pixels that can't be coloured as requested). I think the best way to handle it is to find the closest replacement in perceptual space (look up L*A*B). Or else you could build the error into the dither algorithm. |
