| 01:26.58 | tobiasBora | Hum, so after some time I managed to make the firsts answer work (the one that takes an array of ARGB data), and it kind of work, but it has a very bad gradient (I tried to display a gradient on screen), as it is "cut" in something like 10 colors (and when I was using desc.pixelformat = DSPF_LUT8 it was still worse) |
| 01:27.58 | tobiasBora | However the ereader can definitely do better as it display nicely the gradients when I open a pdf |
| 01:28.12 | tobiasBora | any idea why I just have like 10 colors? |
| 01:29.19 | tobiasBora | (14 to be more precise) |
| 01:44.41 | L29Ah | tobiasBora: fbi(1) can show pictures on fb |
| 01:45.09 | L29Ah | directfb seems to be unmaintained also |
| 01:45.33 | L29Ah | so consider fb/drm/X/wayland for your developments |
| 01:46.23 | L29Ah | also you probably expect dithering that doesn't happen |
| 01:46.57 | tobiasBora | L29Ah: I'm hacking on an existing ereader device that uses directfb, do you think I can use other libs? |
| 01:47.45 | tobiasBora | and I could try to setup fbi to see what it gives... But X/wayland is way too heavy for an ereader, isn't it? |
| 01:47.53 | L29Ah | why not? X can output to a framebuffer device |
| 01:48.21 | L29Ah | you decide; Nokia N810 that's about 15 years old phone, uses X |
| 01:49.42 | L29Ah | not sure if it can do dithering by itself |
| 01:49.54 | tobiasBora | L29Ah: and can it work in parallel with the actual system? I want to keep the actual system, just run from time to time some programs |
| 01:50.13 | L29Ah | dunno what's the system you're talking about |
| 01:50.19 | L29Ah | it grabs fbdev for itself |
| 01:50.26 | tobiasBora | dithering? |
| 01:50.31 | L29Ah | you can use it for other stuff after you kill X |
| 01:50.45 | L29Ah | yes. |
| 01:52.36 | tobiasBora | L29Ah: "it grabs fbdev for itself" -> you talk about X? |
| 01:52.40 | L29Ah | yes |
| 01:53.26 | tobiasBora | ok, maybe I should try... Is there some well explained tuto to setup X? I'm very new to that kind of thing |
| 01:54.12 | L29Ah | dunno, X is like the most common way to do graphics in GNU/Linux so tutorials are all over the place |
| 01:54.17 | tobiasBora | but I'm not sure he will be able to solve my poor gradient... |
| 01:54.46 | L29Ah | consider an image viewer that can do dithering or preprocess your images |
| 01:55.11 | L29Ah | no idea what's the accepted way to do it in software |
| 01:55.40 | tobiasBora | L29Ah: I think I don't understand that. If I send to fb a "perfect" gradient picture, what can go wrong? |
| 01:56.06 | L29Ah | oh, seems like fbi can do dithering |
| 01:56.39 | L29Ah | tobiasBora: nothing wrong, you get what you sent, no dithering |
| 01:57.11 | tobiasBora | L29Ah: I don't know exactly what is dithering. After looking at wiki page, it adds some noise, but how can it help here? |
| 01:58.00 | L29Ah | have you tried to read more than the first sentence? |
| 01:58.07 | L29Ah | what about looking at the supplied pictures? |
| 02:01.03 | tobiasBora | oh I got it, I was on the french page concerning only audio and it was much less clear. So you think that the reader may apply dithering... Interesting... |
| 02:02.46 | tobiasBora | But any idea why when I change the pixelformat between DSPF_LUT8 and DSPF_ARGB the gradient is much better? (From ~5 colors to 14) |
| 02:05.06 | L29Ah | λ> 2 ** (8/3) < 2 ** 4 |
| 02:05.08 | L29Ah | True |
| 02:05.43 | L29Ah | your display stack doesn't understand your display is grayscale and shoves 8-bit coloured bits at it |
| 02:07.27 | L29Ah | or 12-bit in the second case if you mean ARGB4444 |
| 02:08.17 | L29Ah | why haven't you used RGB24? this is an equivalent to 8-bit grayscale |
| 02:09.20 | tobiasBora | hum... but LUT8 is supposed to be grayscale, so I would expect LUT8 to have a better output. But even 8 bits is more than enough to store 5 colors... |
| 02:09.46 | tobiasBora | I used ARGB because the project I was inspired by was using ARGB |
| 02:10.13 | tobiasBora | then I say that the device was turing back to LUT8 automatically when my script was terminating, to I tried LUT8 |
| 02:10.21 | tobiasBora | but can try RGB24 as well |
| 02:11.05 | L29Ah | from the name i expect LUT8 to be a 8-bit lookup table |
| 02:11.20 | L29Ah | make the table containing the "colors" you need |
| 02:11.38 | L29Ah | or maybe the table is in your display, dunno |
| 02:12.03 | tobiasBora | oh maybe you are right... |
| 02:12.42 | tobiasBora | But what happens when I BLIT a ARGB picture on LUT8 if it uses a table? |
| 02:16.37 | tobiasBora | maybe the conversion process explains the strange gradient... |
| 02:16.55 | tobiasBora | anyway, I think I had enough materials to continue to explore. Thanks a lot! |
| 10:52.13 | tobiasBora | L29Ah: Just to let you know, LUT8 + palette + dithering gives a great result, thanks a lot for pointing me out to that direction! |
| 15:18.17 | *** join/#elinux mauz555 (~mauz555@2a01:e0a:56d:9090:11a8:67ae:75b1:26a5) |