Posts: 476
Threads: 25
Joined: Nov 2022
Reputation:
45
08-28-2023, 12:14 AM
(This post was last modified: 08-28-2023, 12:16 AM by grymmjack.
Edit Reason: Spamming the devs with wishes
)
Back to the `_SMOOTH` topic.
That algorithm is fine for photographic elements but it looks bad (IMO) for pixel art.
However! There are anti-aliasing (I think they might be called shaders) algorithms we could also implement into QB64PE - (well you smart people could).
https://en.wikipedia.org/wiki/Pixel-art_...algorithms
It would be awesome to have these as options alternatively to `_SMOOTH`
More stuff:
https://www.reddit.com/r/programming/com...lgorithms/
Repos with code and examples:
https://github.com/search?q=pixel%20art%...positories
C example *cough*: @mkilgore @a740g @smcneill @rhosigma
https://github.com/Treeki/libxbr-standalone
Anyway, if you wanted to try to implement this stuff I would be grateful!
Posts: 3,972
Threads: 177
Joined: Apr 2022
Reputation:
219
08-28-2023, 01:33 AM
(This post was last modified: 08-28-2023, 01:39 AM by bplus.)
i -1 was about the letter indexes for the string. I am sure to keep the fist point drawn at x you have to subtract 1 from i to make a zero offset from x otherwise you will be a whole column right of where you need that first draw point.
Quote:- L16 - The reason we use
-1
in
COLS * scale% -1
and
ROWS * scale% - 1
is because in the first iteration of the loop with
i%
being 1, if we did not subtract one, it would offset them, and so the passed in desired positions would be actually offset for the first character.
Copy / paste above has odd format above,
Anyway, honestly maybe the -1's might not be necessary. Mathologically sometimes you need a -1 and sometimes you don't. A screen with 0,0 top, left corner has bottom right corner at _width -1, _Height -1
If x, y is top, left corner add width, height for bottom right or add width -1, height -1?
Best way to tell is try both. I just tried with the -1's and didn't see any problems.
I think our letter boxes would overlap without the -1's. That is if we outlined the space each letter resides in as boxes, the boxes would overlap without the -1's, whereas the boxes would neatly abut one another with the -1's.
b = b + ...
Posts: 372
Threads: 23
Joined: May 2022
Reputation:
56
08-28-2023, 07:49 AM
(This post was last modified: 08-28-2023, 07:50 AM by a740g.)
(08-28-2023, 12:14 AM)grymmjack Wrote: Back to the `_SMOOTH` topic.
That algorithm is fine for photographic elements but it looks bad (IMO) for pixel art.
However! There are anti-aliasing (I think they might be called shaders) algorithms we could also implement into QB64PE - (well you smart people could).
https://en.wikipedia.org/wiki/Pixel-art_...algorithms
It would be awesome to have these as options alternatively to `_SMOOTH`
More stuff:
https://www.reddit.com/r/programming/com...lgorithms/
Repos with code and examples:
https://github.com/search?q=pixel%20art%...positories
C example *cough*: @mkilgore @a740g @smcneill @rhosigma
https://github.com/Treeki/libxbr-standalone
Anyway, if you wanted to try to implement this stuff I would be grateful!
I love the xBRZ algo. First saw it in action in GZDoom. However, I do not think these algos can be used in real-time unless it is implemented as GPU shaders. GZDoom uses it during the asset loading stage. Which means the textures are pre-processed before any kind of rendering happens.
Posts: 476
Threads: 25
Joined: Nov 2022
Reputation:
45
(08-28-2023, 07:49 AM)a740g Wrote: (08-28-2023, 12:14 AM)grymmjack Wrote: Back to the `_SMOOTH` topic.
That algorithm is fine for photographic elements but it looks bad (IMO) for pixel art.
However! There are anti-aliasing (I think they might be called shaders) algorithms we could also implement into QB64PE - (well you smart people could).
https://en.wikipedia.org/wiki/Pixel-art_...algorithms
It would be awesome to have these as options alternatively to `_SMOOTH`
More stuff:
https://www.reddit.com/r/programming/com...lgorithms/
Repos with code and examples:
https://github.com/search?q=pixel%20art%...positories
C example *cough*: @mkilgore @a740g @smcneill @rhosigma
https://github.com/Treeki/libxbr-standalone
Anyway, if you wanted to try to implement this stuff I would be grateful!
I love the xBRZ algo. First saw it in action in GZDoom. However, I do not think these algos can be used in real-time unless it is implemented as GPU shaders. GZDoom uses it during the asset loading stage. Which means the textures are pre-processed before any kind of rendering happens. Does _PUTIMAGE _SMOOTH do it in advance too? Would imagine yes, like pre-processing?
Posts: 476
Threads: 25
Joined: Nov 2022
Reputation:
45
08-28-2023, 11:51 AM
(This post was last modified: 08-28-2023, 11:55 AM by grymmjack.)
(08-28-2023, 07:49 AM)a740g Wrote: (08-28-2023, 12:14 AM)grymmjack Wrote: Back to the `_SMOOTH` topic.
That algorithm is fine for photographic elements but it looks bad (IMO) for pixel art.
However! There are anti-aliasing (I think they might be called shaders) algorithms we could also implement into QB64PE - (well you smart people could).
https://en.wikipedia.org/wiki/Pixel-art_...algorithms
It would be awesome to have these as options alternatively to `_SMOOTH`
More stuff:
https://www.reddit.com/r/programming/com...lgorithms/
Repos with code and examples:
https://github.com/search?q=pixel%20art%...positories
C example *cough*: @mkilgore @a740g @smcneill @rhosigma
https://github.com/Treeki/libxbr-standalone
Anyway, if you wanted to try to implement this stuff I would be grateful!
I love the xBRZ algo. First saw it in action in GZDoom. However, I do not think these algos can be used in real-time unless it is implemented as GPU shaders. GZDoom uses it during the asset loading stage. Which means the textures are pre-processed before any kind of rendering happens. C pixel scalers here
https://github.com/janert/pixelscalers
XBR guide here @a740g https://github.com/janert/pixelscalers/b...cs/xbr.pdf
Posts: 3,972
Threads: 177
Joined: Apr 2022
Reputation:
219
(08-28-2023, 01:33 AM)bplus Wrote: i -1 was about the letter indexes for the string. I am sure to keep the fist point drawn at x you have to subtract 1 from i to make a zero offset from x otherwise you will be a whole column right of where you need that first draw point.
Quote:- L16 - The reason we use
-1
in
COLS * scale% -1
and
ROWS * scale% - 1
is because in the first iteration of the loop with
i%
being 1, if we did not subtract one, it would offset them, and so the passed in desired positions would be actually offset for the first character.
Copy / paste above has odd format above,
Anyway, honestly maybe the -1's might not be necessary. Mathologically sometimes you need a -1 and sometimes you don't. A screen with 0,0 top, left corner has bottom right corner at _width -1, _Height -1
If x, y is top, left corner add width, height for bottom right or add width -1, height -1?
Best way to tell is try both. I just tried with the -1's and didn't see any problems.
I think our letter boxes would overlap without the -1's. That is if we outlined the space each letter resides in as boxes, the boxes would overlap without the -1's, whereas the boxes would neatly abut one another with the -1's.
Aha! @grymmjack this proves the -1's do neatly place the box spaces next to each other without overlap.
https://qb64phoenix.com/forum/showthread...6#pid19096
I fixed my own "Font Print" sub by adding -1's to the Step when drawing the pixelated boxes that form a letter.
Notice how you can see each pixel individually specially at the bottom b+ with 5x5 font.
b = b + ...
Posts: 476
Threads: 25
Joined: Nov 2022
Reputation:
45
(08-29-2023, 03:21 PM)bplus Wrote: (08-28-2023, 01:33 AM)bplus Wrote: i -1 was about the letter indexes for the string. I am sure to keep the fist point drawn at x you have to subtract 1 from i to make a zero offset from x otherwise you will be a whole column right of where you need that first draw point.
Quote:- L16 - The reason we use
-1
in
COLS * scale% -1
and
ROWS * scale% - 1
is because in the first iteration of the loop with
i%
being 1, if we did not subtract one, it would offset them, and so the passed in desired positions would be actually offset for the first character.
Copy / paste above has odd format above,
Anyway, honestly maybe the -1's might not be necessary. Mathologically sometimes you need a -1 and sometimes you don't. A screen with 0,0 top, left corner has bottom right corner at _width -1, _Height -1
If x, y is top, left corner add width, height for bottom right or add width -1, height -1?
Best way to tell is try both. I just tried with the -1's and didn't see any problems.
I think our letter boxes would overlap without the -1's. That is if we outlined the space each letter resides in as boxes, the boxes would overlap without the -1's, whereas the boxes would neatly abut one another with the -1's.
Aha! @grymmjack this proves the -1's do neatly place the box spaces next to each other without overlap.
https://qb64phoenix.com/forum/showthread...6#pid19096
I fixed my own "Font Print" sub by adding -1's to the Step when drawing the pixelated boxes that form a letter.
Notice how you can see each pixel individually specially at the bottom b+ with 5x5 font. Yeah I noticed a big difference. Your second attempt is a lot cleaner! Well done.
|