• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

DF Retro: The Super Nes Classic Mini Review

I’m not really a fan of razor sharp scan lines so the CRT filter ain’t bad for me. Wish it was more color accurate but I think it has a positive impact on the mode 7 games, for the rest I’ll stick to the 4:3.

Would be curious as to how it compares to other emulation boxes, things like retropi and retron 5 to me is more interesting than original hardware since it’s closer to the price of the SNES mini.
 
No, it’s still not there. The Mini output looks as if it’s using bilinear filtering which is not what you get with a consumer CRT. I think Sonic Mania’s two CRT options are an example of how to do it right.


This! I never, ever liked such filters but Sonic Mania did them so fantastic, I was unbelievable impressed, specially from the last one where it really looks like you are looking at an original Game Gear screen in handheld mode. Just amazing.
I don't like the SNES mini mode cause it's just blurry, looks awful and has lines in between. It gives me personally nothing and only makes the quality of the picture worse, it's just technically bad implemented.

If I want something like that I connect my SNES with old standard AV cables. (not RGB cables they were awesome, the other ones)
 
Also DF thank you for confirming what I suspected all along: that the console does has 1-2 frames of input lag compared to real hardware.
 
I don't think the CRT filter offered here is accurate in any way. Like dark10x said, it's just a bilinear filter plus scanlines. It's missing a lot of the uh grime that playing on composite would give you, such as color bleeding. In a perfect world, this little box would have advanced settings that lets you tweak it however you want.

tn1ycku.jpg


The Sony CRT in the video is obviously not how anyone looked at the games back then though, not even the developers! :P I maintain these games were designed to be played with non-perfect pictures.

What game/software is that option box from?
 
Oh thank god. Was about to cancel my pre-order.

Btw, dark. Do you edit these videos yourself? And how do you manage to get identical video footage from games on different systems? (i.e. the MMX comparisons)
Yes, I do 100% of the production on these videos. For matched footage, I actually just used the attract mode demo on the game!
 
Yeah, I'd be curious to hear what people consider the best CRT shaders and the SNES emulators they're compatible with. I recently got into Higan but it seems to be less capable on the shader front when compared to other emulators.
 
Yes, I do 100% of the production on these videos. For matched footage, I actually just used the attract mode demo on the game!

Wow. Well, that's some good stuff.

Mind if I ask what editing program you use?
 
Yeah, I'd be curious to hear what people consider the best CRT shaders and the SNES emulators they're compatible with. I recently got into Higan but it seems to be less capable on the shader front when compared to other emulators.

http://emulation.gametechwiki.com/index.php/CRT-Royale

It's uber-customisable, Kurozumi's version apes the old sony reference crt's I believe

https://github.com/libretro/common-shaders/blob/master/cgp/crt-royale-kurozumi.cgp

I think byuu and themaister were working out a new bsnes core for retroarch, so I imagine that combo would be best.
 
Very surprised to see that there was no discussion of the Super NES Classic running games slower than the real hardware, only a comment that Star Fox 2 runs slowly - and not that it runs slower than the real thing.
Here's a comparison video with the original Star Fox: https://www.youtube.com/watch?v=-YJLGwwITv8
Looks like it could have other timing-based errors as well, since the level played out exactly the same on both consoles, but differently on the SNES Classic.
Other games appear to be affected as well based on the footage I've seen from the SNES Classic, but I haven't seen people set up direct tests for comparison. I was expecting Digital Foundry to be the ones to do that.

It looks like the OSSC is not performing colorspace conversion, and is outputting the original BT.601 color - either that or it's an editing/capture issue.
Video as played on YouTube in BT.709:
709i8srk.jpg

Dull greens typical of incorrect color decoding.

Video played with the input matrix set to BT.601:
601q8sjw.jpg


It's a much closer match to emulation:
emulatedbhsq3.png
Note: both captures were brightened up a little as they were very dull.
White was approximately 218 which suggests another levels conversion issue. (double level compression turns 0-255 into 30-218)

Great video, though I do disagree on the CRT filter verdict.
Yeah, it's blurry, but so were probably 90% of the actual TVs people had back then. Most people didn't have those Sony CRTs
A lot of it depends on the CRT you're trying to emulate, and a professional video monitor is not representative of a consumer-grade CRT at all - not even a Trinitron.
I don't necessarily agree that CRT filter should be blurry, but it should have a much lower horizontal resolution.
A professional monitor might have something in the region of 600-800 TVL while a consumer CRT may be 250-300 TVL.

Here's what a 560 TVL PVM looks like compared against an 800 TVL PVM:
The higher resolution the CRT, the more defined the scanlines are, and horizontal lines start looking more like unbroken lines than being drawn from a grid of pixels.
I still think that Royale is the best CRT filter - though it's demanding to run and requires a lot of tweaking to look good so it would never be used on something like this. It can be configured to emulate most types of CRT, including a BVM. (not my shot)
 
Wow. Well, that's some good stuff.

Mind if I ask what editing program you use?
I use Adobe software for everything. Works pretty well.

Other games appear to be affected as well based on the footage I've seen from the SNES Classic, but I haven't seen people set up direct tests for comparison. I was expecting Digital Foundry to be the ones to do that.
I mentioned it in regards to Kirby which runs faster.
 
My brief time with both my retropie (running retroarch/snes9x) and the SNES mini suggests that the SNES mini is a bit less laggy. Anyone else agree?

I know you can reduce the frame rate with the dispmanx gpu option but the SNES mini still feels a bit less laggy to me.

I know the retropie OS stack is inherently a little slower than PC emulation. I thought the SNES mini was surprisingly unlaggy.

Great performance review, I love the DF Retro videos!
 
http://emulation.gametechwiki.com/index.php/CRT-Royale

It's uber-customisable, Kurozumi's version apes the old sony reference crt's I believe

https://github.com/libretro/common-shaders/blob/master/cgp/crt-royale-kurozumi.cgp

I think byuu and themaister were working out a new bsnes core for retroarch, so I imagine that combo would be best.

Thanks. I'm familiar with the shader although I'm not very good at configuring it.

A quick search isn't showing me exactly which RetroArch core I should use for Byuu's work. It sounds like a Higan core will be added but I don't think it's happened yet.
 
They did comparisons with alternates? Did they compared it with the Retron5? (Have to watch it later)
This is still the ne plus ultra for me. 5 cartridge slots, you can use your original controllers, same 1080p razor sharp pixel quality as with the SNES classic you use your original savegame, save it on the machine and back to the cartridge whenever you want. It's a dream for collectors and guys who wants to play NES, SNES, Genesis, and Gameboy/GBA games in the best quality possible. I love my SNES classic, even if I don't wanted to get one first, but nothing stands a chance against the Retron5 in my opinion. Still the best one out there.

The retron5 has more emulation issues and has horrendous input lag compared to real hardware, or even the SNESCE. This also sounds so much like an advertisement because it's definitely not the best quality possible, especially with sound accuracy. Plus the whole theft of other people's hard work that I don't think they ever addressed...


As for the video itself, I take offense to the the concept of Mario RPG and Kirby SS not taking advantage of the SA-1. Just because they didn't pump polygons everywhere didn't mean they weren't making great use of the chips additional ram and processing power. You'll note that nearly everything in KSS uses very large very detailed sprite work, this is possible only because of the additional ram/vram for example. Also note, those demos you showed by homebrew developers are impressive, but you wouldn't be able to fit any gameplay logic in using them for something so complex. They are the epitome of tech demos.
 
Oh I missed that. Still, a lot of games appear to have more slowdown or worse slowdown than the real hardware.

I think that’s hard to prove since it’s really hard to get the same instance on both the SNES classic and original hardware.

Slowdown was fairly prevelant on the original SNES.
 
Oh I missed that. Still, a lot of games appear to have more slowdown or worse slowdown than the real hardware.
Which ones? I didn't notice that at all when doing side by sides. The footage mostly lined up except for Kirby which always finished sections faster on the Mini.

There are certainly going to be minor timing differences, though, but it should be subtle enough not to cause severe issues. That's just the nature of non-cycle accurate emulation.
 
Fantastic work as usual, John. Great summary of the hardware differences. I'm thinking I may finally get in on the next batch of OSSC orders.
 
The retron5 has more emulation issues and has horrendous input lag compared to real hardware, or even the SNESCE. This also sounds so much like an advertisement because it's definitely not the best quality possible, especially with sound accuracy. Plus the whole theft of other people's hard work that I don't think they ever addressed...

No idea what you mean, I know the Retron5 for years now and there is nothing that comes close to it if you want to use your cartridges and have the picture in 1080p. With all the patches, the compatibility isn't that bad anymore.
What would be a better alternate with the exact same features?
 
Forgive my ignorance but couldn't emulating the snes games on much more powerful hardware fix the slowdown.
Well, it depends. With a cycle accurate emulator, which requires a beefy machine, it won't fix it because it's designed to simulate the console 100% accurately.

Some emulators can speed things up slightly, though.
 
Well, it depends. With a cycle accurate emulator, which requires a beefy machine, it won't fix it because it's designed to simulate the console 100% accurately.

Some emulators can speed things up slightly, though.
Thanks that makes sense.
The anti epilepsy stuff is weird.
 
Thanks that makes sense.
The anti epilepsy stuff is weird.

It wouldn't require a beefy machine at all. In fact, early SNES emulators you could 'overclock' the SNES. Though I believe it was removed in later versions, Snes9x used to have a command that was '-h xxx' where xxx was a percentage of the clockspeed. Messing with this would cause all sorts of issues with some games, but it could be used to remove slow downs.
 
Forgive my ignorance but couldn't emulating the snes games on much more powerful hardware fix the slowdown.

Well, it depends. With a cycle accurate emulator, which requires a beefy machine, it won't fix it because it's designed to simulate the console 100% accurately.

Some emulators can speed things up slightly, though.

In most cases this wouldn't be possible, however it is possible to overclock the SuperFX chips to run Star Fox/Star Fox 2 more smoothly. One of the mercury forks of the BSNES core in Retroarch supports this.

Star Fox overclocked
Star Fox 2 overclocked
 
Is it more laggy than retroarch with gpu sync on?

Never tried it, no idea. Despite the lag, it ranks up there with the Wii VC as the most responsive ones I’ve ever played. I tried running Wii homebrew SNES/NES emulators (FCE Ultra GX, etc.) and they all had more lag than this.

The only way I ever got to feel like an emulator had zero input lag was on PC and turning off V-Sync, meaning I was stuck with horrid screen tearing...
 
I heard Terranigma underworld music at one point in the video. Excellent taste!

Although I'm from the US and Terranigma was never officially released here. I got the chance to play it on real hardware years ago.
 
I heard Terranigma underworld music at one point in the video. Excellent taste!

Although I'm from the US and Terranigma was never officially released here. I got the chance to play it on real hardware years ago.

Yup, definitely Terranigma. What a kick in the nuts to hear that music play. Knowing we'll never see a re-release of this game sucks.
 
These overclocks increase the framerate but they also increase the speed of the game. I wonder if Starfox was designed with the slowdown in mind.

Probably because the audio doesn't fully play during the cutscene esque segments and is cutout before it finishes.

---------------------------------------------------------------------------------------------------------------------

In regards to modding the NES classic. Was it possible to outright replace the stock roms included with the unit? Like using a newer revision of a rom to replace an older one that was included in it for the sake of having bug fixes or even the removal of the modified dulled flashes for the sake of epilepsy protection?

If this is possible. Then it might be worthwhile to do this on the SNES classic too such as replacing the stock Contra III and Yoshi Island roms to revert the changes made such as the flash protection and disabled effects.
 
Probably because the audio doesn't fully play during the cutscene esque segments and is cutout before it finishes.

---------------------------------------------------------------------------------------------------------------------

In regards to modding the NES classic. Was it possible to outright replace the stock roms included with the unit? Like using a newer revision of a rom to replace an older one that was included in it for the sake of having bug fixes or even the removal of the modified dulled flashes for the sake of epilepsy protection?

If this is possible. Then it might be worthwhile to do this on the SNES classic too such as replacing the stock Contra III and Yoshi Island roms to revert the changes made such as the flash protection and disabled effects.

No, they could only be hidden. So you'd hide the old one and upload another one.

Epilepsy filter was a global thing though, and could be disabled.

If Yoshi is a filter, that will fix it. If not, it would have to be pointed at retroarch, but I don't think the onboard games can be configured like that (or at least the nes ones weren't), so again, you'd upload another and point it at retroarch to run.
 
The only way I ever got to feel like an emulator had zero input lag was on PC and turning off V-Sync, meaning I was stuck with horrid screen tearing...
I know what you mean...

RetroArch fixed those problems though. I don't have a SNES mini to compare unfortunately.
 
I know what you mean...

RetroArch fixed those problems though. I don't have a SNES mini to compare unfortunately.

What is RetroArch? Doesn't it just downloads other emulators and plays them through their own overly complicated UI? Or is it a emulator built ground up?
 
What is RetroArch? Doesn't it just downloads other emulators and plays them through their own overly complicated UI? Or is it a emulator built ground up?
It's a frontend like program that runs it's own emulation cores. That means existing emulators need to be ported as cores first but after that they all share the same interface. It's like Open Emu on Mac but it has more capabilities.

It might be complicated at first but i assure you, it's worth it when you get used to it. You get perfect sound/video sync, options to lower input lag and the best CRT shaders. And that OG gameboy shader, jesus...
 
John talks about the 8bitdo SNES/SFC controllers occasionally recognizing unintended down presses on the d-pad. Has anyone tried a scotch tape fix a la Switch Pro controller?

Very curious about this.
 
Top Bottom