Framework Update?

[justaguy]

The last release for the TurboGrafx core was in March, and since then pixel repetition support has been added to the framework. This is a feature I'd really like to have working in this core because it allows for integer scaling with correct pixel aspect ratio (8:7 for low-res, 6:7 for mid-res, 4:7 for high-res) which I've been enjoying on the NES and SNES cores. Would it be too much trouble for someone to update the framework and do a release? If it's just a matter of replacing the sys folder, compiling, and doing a quick smoke test I could probably submit a pull request myself, but I thought I should ask first since an actual FPGA dev could probably get it done more efficiently.

Update: I see there's already a pull request open for this. Thanks makigumo!

[Retro-Nerd]

I second that. My C9 can't use a 1920x1440p resolution, so supporting the pixel repition 2560x1440 resolution would be nice for the TurboGrafx/PC-Engine core. The integer scaling results are much better than using 1920x1080p.

[justaguy]

I second that. My C9 can't use a 1920x1440p resolution, so supporting the pixel repition 2560x1440 resolution would be nice for the TurboGrafx/PC-Engine core. The integer scaling results are much better than using 1920x1080p.

Unfortunately, pixel repetition won't actually give you 2560 pixels of horizontal resolution, but rather 1280 double-wide pixels, so your desired horizontal scale factor has to be an even number. If you don't mind having square pixels, you can do 6x6, but if you want a close-to-accurate pixel aspect ratio, 1080p with the 216p crop option gets you closer (6:5 for low-res, 4:5 for mid-res, 3:5 for high-res--all less than 10% off).

[Retro-Nerd]

I don't like the cropping, it feels like a small zoom too. It isn't a 4:3 display aspect anymore, which i prefer. Also there are some games that show something important at the most top (lives counter, score etc). I know that the pixel repititon is actually a 1280x1440p output, but it looks good enough for me. I use CRT filters anyway.

The PR 1440p maybe doesn't give me a full screen for PCE games, but it should integer scale a bit larger than 1080p.

[justaguy]

So after doing some testing with the core from makigumo's PR and the current official release, it looks like pixel repetition works fine without a framework update, although there is some oddness with horizontal filtering. I'm not sure why it wasn't working for me before. Maybe I had a bad modeline or the resolution was too high for the 1080p display I was testing on at the time to accept (NES and SNES worked fine with only slightly lower resolutions though). Anyway, if you want to try it out, the modeline I used is:

Code: Select all

video_mode=1080,24,16,40,1617,3,10,33,115625,+hsync,-vsync,pr

I used the CVT-RB timings from this calculator for 2160x1617 at 60 Hz, dividing the H values--i.e. the first four numbers--and pixel clock by two.

(It looks like Sorgelig doesn't want people submitting pull requests that include releases, so we'll have to wait on him for a new release with the updated framework.)

[Retro-Nerd]

This isn't working for me. I'll wait for an update. Thanks for posting the modeline though.

[Chris23235]

This isn't working for me. I'll wait for an update. Thanks for posting the modeline though.

Sorgelig just updated the framework of the core.

[Retro-Nerd]

Yep, everthing's fine now. Thanks for the update.