Re: Pictoric (was: New conversion algorithm)
Posted: Mon Sep 30, 2019 11:21 pm
The comparison of Dali's painting you made, shows many differences
That's odd. Removing the pixels shouln't alter the remaining ones. But your 2 versions of Dali's paining isn't showing this.
If I analyze the 1st picture above, I dont count 8 unique colors but 256. How is this possible? An oric picture should have exactly 8 colors max. In addition, it seem the aspect ratio isn't respected on the 2nd picture. Did you convert some modified version of the ".tap.bmp" file ?
[EDIT] oh! I know hat you did which explains this: you captured the oric picture from an oric emulator. Mis mean that your 2nd source isn't 240x200 but includes extra black pixels in the bottom of the image (this is especially visible on the "cleaned" version above: the bottom part has now 4 times more black rows). The convesion will then blend the dithered pixels reducing quality alot. You should'nt need to do that. You should use the ".tap.bmp" file as 2nd source. It contains the exact content of the tap file, but in bitmap version for easy hand-cleanup. Since the size is correct, no blending occurs and the remaining pixels are left untouched.
Here is how I would "improve" a picture after conversion by hand. First I'd do the conversion in grafX2. Then once the converted oric picture is displayed, I'd remove the "noisy" pixels by hand, and call the convertion once again. As you can see, there is no changed except for the removed pixels.
Maybe in v1.1 I'll add code to consider that low-intensity pixels are actually black, removing these scatered pixels here and there. But I need to test before, because this could also alter the quality of some other pictures.
That's odd. Removing the pixels shouln't alter the remaining ones. But your 2 versions of Dali's paining isn't showing this.
If I analyze the 1st picture above, I dont count 8 unique colors but 256. How is this possible? An oric picture should have exactly 8 colors max. In addition, it seem the aspect ratio isn't respected on the 2nd picture. Did you convert some modified version of the ".tap.bmp" file ?
[EDIT] oh! I know hat you did which explains this: you captured the oric picture from an oric emulator. Mis mean that your 2nd source isn't 240x200 but includes extra black pixels in the bottom of the image (this is especially visible on the "cleaned" version above: the bottom part has now 4 times more black rows). The convesion will then blend the dithered pixels reducing quality alot. You should'nt need to do that. You should use the ".tap.bmp" file as 2nd source. It contains the exact content of the tap file, but in bitmap version for easy hand-cleanup. Since the size is correct, no blending occurs and the remaining pixels are left untouched.
Here is how I would "improve" a picture after conversion by hand. First I'd do the conversion in grafX2. Then once the converted oric picture is displayed, I'd remove the "noisy" pixels by hand, and call the convertion once again. As you can see, there is no changed except for the removed pixels.
Maybe in v1.1 I'll add code to consider that low-intensity pixels are actually black, removing these scatered pixels here and there. But I need to test before, because this could also alter the quality of some other pictures.