Missing fit scaling alternative in cli version

[magwed]

Thanks for a great tool and for sharing!
I have an idea of setting up a hosted Web API to be able to automatically convert images on the fly from C64 itself.
Does the project include a Web API currently?
Your CLI tool seem to be the perfect start of such a project.
I saw that the very useful "fit" scaling property avaiable in the web was missing from CLI and would be most useful to have included.
Also a more full control over JIMP would be cool, like a simple way to control: Contrast, Brightness, Saturation and other parameters from CLI.

[micheldebree]

Hi @magwed. It would be good to add the "fit" functionality aswell to put it in line with the web version. I am a bit reluctant to add more pre-processing to the cli version because:

• It is meant as a developers tool as part of an (automated) conversion process. A pre-processed image can be fed into the cli, and the cli focusses on it's main task
• I might want to swap out Jimp for another library as it is giving me headaches in the web version. The pre-processing in the web version is all Jimp, and I do not want to be tied so much to Jimp in the core and cli versions.

It should not be very hard to do some pre-processing in your API, with Jimp or otherwise, it's a separate step before using the cli.

There is no Web API currently. I have been toying with the idea. The web version now runs 100% on the client browser, but separating it into GUI and API would also help with the GUI headaches :)

I am not sure when I can add the 'fit' functionality, but you could check the web sourcecode how it's done and implement it as a pre-processing step.

Instead of using the cli, you could also use the core component. The cli and web version both use that.

[mwedmark]

Thanks for the comment. Yea I guess I can add this myself to my own fork and you can do as you like.

Just for feedback your retropixels is now being used directly from C64 using the open-source project meatloaf which makes it possible to access internet using a virtual IEC-device. By using a special syntax, I'm using retropixels to convert a http-linked image into a executable koala-image, returning and loading it into C64 memory. From a user-perspective you pretty much "load" the JPG-file and show it on-screen. Nice huh?

[micheldebree]

That's very nice to hear, that retropixels is actually being used!

My suggestion is not to fork retropixels, but to pre-process the image before feeding it into the cli. The reason I want to keep the pre-processing options in the cli to a minimum is that pre-processing like scaling, color manipulation etc. is a separate step from the actual conversion. By keeping it separate you can have control over the pre-processing without being dependent of what options retropixels has to offer. At the same time, the effort put into retropixels can focus on actual conversion instead of generic image manipulation.

I plan to add the 'fit' option to the cli soon because I agree it is not "symetrical" with the web version options.

[micheldebree]

I have pushed 0.9.10 of the cli version (retropixels), with the added 'fit' option for --scale. Could you give it a spin?

[mwedmark]

I tried it and it looks perfect! Portrait images fits right into the landscape C64 original format, great quick fix! I like it!

Just one question, what happened to the "-f prg" that I used for my server?
I'm doing this and gettting the below error:
retropixels.cmd .\panda-standing.png --overwrite -f prg -o panda.prg -s fit

-snip-
C:\Users\Magnus\AppData\Roaming\npm\node_modules\retropixels\cli.js:138
throw new Error(Executable format is not supported for ${binary.formatName})
^

Error: Executable format is not supported for Koala
at saveExecutable (C:\Users\Magnus\AppData\Roaming\npm\node_modules\retropixels\cli.js:138:11)
at C:\Users\Magnus\AppData\Roaming\npm\node_modules\retropixels\cli.js:175:7
-snip-

[micheldebree]

Sorry something seems to have broken along the way. If you upgrade to the latest version (0.9.11), it should be fixed.

[mwedmark]

Thanks! Now all works for me. Thank you, great work!