Hey everyone!
So, we are currently looking very seriously at a console release for Defender's Quest II. Of course, PC is our top priority, and don't take this as an official statement of specific platform choices or anything just yet, we still have a lot of details to work out.
I think the upscaled art looks fine. I liked the original pixely aesthetic, but the new one makes the interface look better (the “Battle” button looks out of place in the pixel version). You’ll want to touch up anything that’s supposed to be square when you upscale it though, since everything got kind of rounded (especially noticable in cities).
Any chance you can re-release DQI for Linux once you do this? I only made it part of the way through since getting Adobe Air installed was too had to do on my main computer (but I played part of it on a Mac).
That would be one of the many, many benefits of having the native Haxe-based DQ engine ready. We could put out a solid, NATIVE, C++/SDL build of Defender’s Quest 1 for linux.
I like the smoothness, you’ve done a good job keeping it faithful to the original yet making it pleasing to the eyes in higher res.
Reminds me of the old Snes emulator filters. http://www.fantasyanime.com/emuhelp/emuhelp_zsnes.htm
I prefer the pixels but it looks better with the filter.
The XBR algorithm was in fact originally created by and for the emulator scene 
It’s quickly become a standard:
This looks very good while still keeping the simple and nice pixelation we’re all used to, I really like it!
Edit: just saw this from my phone, pixelated looked a bit “choppy” but new one adapted well and looked perfect!
That upscaled art looks nice enough even without the editing, though the edited versions sure look more polished.
I’m not surprised that your upscaling algorithms come from emulators, but that also means they need to run in real time. I wonder if there aren’t some options that could do a little more math, but result in a higher quality image or even vector art.
Actually, a quick search turned one up: http://research.microsoft.com/en-us/um/people/kopf/pixelart/
This one shows some comparisons, including a bunch of upscaling options. Potrace output seems interesting. http://research.microsoft.com/en-us/um/people/kopf/pixelart/supplementary/index.html
Hey Peter!
It’s a good point, if there are offline algorithms that are too slow for emulator use, but still produce high quality output they’d definitely still be on the table.
That MS research algorithm was actually the first one I evaluated, and it was my top contender until I compared it against the results of XBR. (XBR was invented a few years after the MS research method was published)
Looks pretty good to me!