• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Digital Foundry vs Mario Kart 8

exfatal

Member
There was once a prince, and he wanted a princess, but then she must be a real Princess. He travelled right around the world to find one, but there was always something wrong. There were plenty of princesses, but whether they were real princesses he had great difficulty in discovering; there was always something which was not quite right about them. So at last he had come home again, and he was very sad because he wanted a real princess so badly.

One evening there was a terrible storm; it thundered and lightninged and the rain poured down in torrents; indeed it was a fearful night.

In the middle of the storm somebody knocked at the town gate, and the old King himself sent to open it.

It was a princess who stood outside, but she was in a terrible state from the rain and the storm. The water streamed out of her hair and her clothes; it ran in at the top of her shoes and out at the heel, but she said that she was a real princess.

‘Well we shall soon see if that is true,’ thought the old Queen, but she said nothing. She went into the closet, dusted off the Wii U and popped in Mario Kart 8. This was where the princess was to game that night. In the morning they asked her how she played.

‘Oh terribly bad!’ said the princess. ‘I could nary finish a single race! Heaven knows what was wrong with the framerate. It seemed to be repeating one out of every sixty-four frames on occasion, and my whole vision is blurry this morning. It is terrible!’

They saw at once that she must be a real princess when she had felt the frame being repeated. Nobody but a real princess could have such a delicate eyesight.

So the prince took her to be his wife, for now he was sure that he had found a real princess, and the repeated frame was put into the Museum, where it may still be seen if no one has stolen it.

Now this is a true story.

beautiful story, should be made into a movie. but on topic DF job is to break down the techy aspect of the game and tell you. Just unfortunate that people will cancel preorders due to this news and miss out on a great game.
 

Kai Dracon

Writing a dinosaur space opera symphony
I suppose it's worth noting that in one Pikmin 3 bug fix patch Nintendo did address slight framerate drops that happened reliably during certain animations in the game.

No way to know if they'll consider this a significant performance issue however. It does seem like an honest bug. Especially since other 60fps Wii U games developed by Nintendo and Retro have no oddities in their framerate.
 

Ty4on

Member
To be fair, it probably IS easier to render 2 640x720 screens than it is 1 1280x720 and 1 (insert Wii U gamepad resolution here) screen.

Seems to be the reason.
I'd love to learn what the cause of the dropped frame every second is. Seems to be something funky about the timing.

Shame about the lack of AA and filtering. No surprise given the HW, but the art style really benefits from smooth IQ.
 

EatChildren

Currently polling second in Australia's federal election (first in the Gold Coast), this feral may one day be your Bogan King.

Caladrius

Member
*In which people don't notice the confirmation that the game runs at 60 online, i.e. the mode you'll be playing the vast majority of the time.*

Is this the kind of thing that can be fixed?

If it actually is a bug/design flaw as opposed to a technical constraint, yes.

The real question is if Nintendo feels it's urgent enough to merit a patch.
 

Wasp

Member
Yeah it's a bug for sure. Hopefully Nintendo will fix it. (and it would be nice if they patched in a bit of anti-aliasing whilst they're at it)

It's crazy that they didn't detect this themselves. I would've thought they would have ran the game through a framerate analysis before it went gold.
 

MilesTeg

Banned
Seems to be the reason.
I'd love to learn what the cause of the dropped frame every second is. Seems to be something funky about the timing.

Shame about the lack of AA and filtering. No surprise given the HW, but the art style really benefits from smooth IQ.

It seems to be a bug with the AI.

Game runs at 60fps in time trial or whatever.

Game runs at 60fps in split screen VS (no AI).

But as soon as you get a CPU controlled player on screen, you get the repeated frame issue.

At least this is how I'm understanding it.
 
One of the first things I noticed when I played the demo was that it wasn't a constant 60fps. Game is still very pretty and smooth enough, though.
 

VICI0US

Member
the 59/60fps fluctuation is basically imperceptible to me in actual gameplay.

couple that with the fact that it's only present in CPU races and i really don't care, 99% of my time will be spent playing online with humans
 

Talamius

Member
To be fair, it probably IS easier to render 2 640x720 screens than it is 1 1280x720 and 1 (insert Wii U gamepad resolution here) screen.

Probably, but I'll miss that feature from Sonic. Giving the Gamepad player their own screen was a nice little touch.
 

Ty4on

Member
It seems to be a bug with the AI.

Game runs at 60fps in time trial or whatever.

Game runs at 60fps in split screen VS (no AI).

But as soon as you get a CPU controlled player on screen, you get the repeated frame issue.

At least this is how I'm understanding it.

It could be a bad value or something, but it could also be that the AI messes something up and fixing it would be too complicated.
 

Thrakier

Member
Oh man, it's as if you plucked my own response this thread straight from mind. I have absolutely zero idea what difference it makes and I never notice these issues when I play games. It must be depressing to not be able to enjoy a game because of such minor issues.

On the other hand I'm not jumping on DF for their findings. This is what they do. They can't sugarcoat impressions if their aim is to produce hard facts.

So, you do know jack shit, you don't see jack shit, you probably weren't even able to tell if the game ran at 25FPS but you know ONE thing for sure: It's a MINOR issue. It's not just something you don't notice because your perception is sub par or you are just used to shit, no. It's not you. It's the OTHERS whose fault that is, because it's minor!!!

Great logic and stuff. Go on.
 

DECK'ARD

The Amiga Brotherhood
Happening every 64th frame doesn't sound like a bug, it sounds like a cheat to reduce CPU load for the AI.

Rather than the AI running constantly it's using the time of the repeated frame to update all the CPU players. Making a decision and giving them a target effectively for the next 64 frames.

The engine is pushing the Wii U harder than previous ones so it's possible they hit a limit. If the AI was running every frame and didn't calculate things in time you'd end up dropping to 30.
 
If it's an internal timing issue that just totally slipped QA I guess they could patch it pretty easily. At least that's what it sounds like.
 

TwistedN

Member
Nintendo is pretty serious about performance, so I imagine it will get fixed at some point. It doesn't seem like something they'd put in patch notes though.
They might patch it being a game built from the ground up for Wii U, but Wind Waker HD's framerate was never dealt with :/
 

dark10x

Digital Foundry pixel pusher
It could be a bad value or something, but it could also be that the AI messes something up and fixing it would be too complicated.
What's interesting, and a detail I didn't mention, is that playing with a ghost (in time trials) does not incur this hit either. It really is limited to active AI racers. If you play online with AI racers in the mix you're going to see this issue. Provided you're playing only with humans you'll be fine.
 
I'll just treat it like I do oxygen, I'm aware its there but I'm never noticing it or actively think about it. I looked at the example posted and didn't notice a thing until I saw it slowed down.
If its consistent then its a feature not a bug!
kidding
.
 
To be fair, it probably IS easier to render 2 640x720 screens than it is 1 1280x720 and 1 (insert Wii U gamepad resolution here) screen.

You cannot play two player mode with one person on the TV and the other on the Gamepad getting a whole screen to themselves ?.

http://a.pomf.se/gdqvuj.webm

Around the 9 to 11 second mark as Mario is rounding the corner, you can see it pretty well in the words "Star Cup" as it goes by.

I can't even see that when looking for it lol...
 

Just lol... and people wonder why some ask for a Gamepad free SKU to bring the price right down, first party developers can't even be bothered to use it in the most simple ways. Sonic All Star Racing had 5 players with one using the Gamepad FFS and it was a launch game...
 

Ty4on

Member
What's interesting, and a detail I didn't mention, is that playing with a ghost (in time trials) does not incur this hit either. It really is limited to active AI racers. If you play online with AI racers in the mix you're going to see this issue. Provided you're playing only with humans you'll be fine.

Does the number not matter? Like only having one AI car (if possible) causes this as well?
SM3DW is way less computationally intensive than Mario Kart 8. Like, I doubt it's even a close contest.
They're probably quite equal given the similar framerate. Perfect Dark Zero is computationally intensive, but that doesn't automatically make it good looking.
 
If only it were pointed out in the article or this thread about 50 times.

I obviously didn't read the entire thread, but I shouldn't have to. I'm just not buying the basis of Digital Foundry's argument here. I think what's more likely to be the case is that this is evidence of how insanely flawed Digital Foundry's method of analysis can truly be. We appear to have a situation where their tools or methods of analysis are telling them one thing, and then they're desperately trying to seek ways to rationalize the data they're getting by exaggerating the perceived impact, despite the fact that it appears the game in action isn't supporting what they're suggesting.

"Notice how the frame-rate inconsistencies manifest during gameplay. This issue is difficult to observe in a 30fps video but quite noticeable during actual gameplay"

They tell us that we should notice how frame-rate inconsistencies manifest during gameplay based on the video they've provided, and then immediately afterwards tell us how difficult to observe the issue is in a 30fps video. So, which is it? The video either proves on some level what they're saying to be true, or it doesn't. Could it be the only thing from that video that even remotely comes close to proving any of the claims they're making is that framerate number that they themselves provide in the upper left hand part of the screen? Forgive me, but it sounds like they've thought up some bs to defend the integrity of their analysis. If it's truly as noticeable as they say, why not go the extra mile and provide us a proper 60fps video that would prove how very noticeable this is? Because, really, with the exception of their own number that they provide in the upper left portion of this video, nothing in the video supports their claim as far as I can see.

I've always believed to a degree that Digital Foundry's framerate tests sometimes felt as if they were questionable or outright flawed in terms of the data they gave us, and how they then claimed that data directly translates to the actual user experience, and I think this article strongly supports that belief. I may very well notice what they claim when I see this game for myself - and I will considering I bought a copy for my sister in law, which is visiting me for a week starting June - but until I see that for myself, I'm calling bogus on their claims as to how noticeable this supposedly is during actual gameplay.
 
They might patch it being a game built from the ground up for Wii U, but Wind Waker HD's framerate was never dealt with :/

Completely different situations. WWHD has an inconsistent frame drop caused from graphic load.

MK8 has consistent frame freezes every 64th frame due to... well... we don't know yet. But we know it's not a graphic issue because it runs perfectly when AI are disabled and in online mode.
 
Why does WiiU have such an issue with AA ?, is it just a design choice to leave it out to hit 60fps on the limited hardware or do the developers simply not care ?.

Pikmin 3 ran at 30fps and still suffered from jaggies and W-101 was riddled with them while regularly dropping below 60fps.

SM3DW and WW HD are the "cleanest" WiiU games I have seen.
 
I obviously didn't read the entire thread, but I shouldn't have to. I'm just not buying the basis of Digital Foundry's argument here. I think what's more likely to be the case is that this is evidence of how insanely flawed Digital Foundry's method of analysis can truly be.

Whatever point you want to make about DF may or may not be valid, but this is a bad article to make your stand on. This is not a subjective issue he has brought up - you can slow down the video and identify the duplicate frame.
 

Thrakier

Member
I obviously didn't read the entire thread, but I shouldn't have to. I'm just not buying the basis of Digital Foundry's argument here. I think what's more likely to be the case is that this is evidence of how insanely flawed Digital Foundry's method of analysis can truly be. We appear to have a situation where their tools or methods of analysis are telling them one thing, and then they're desperately trying to seek ways to rationalize the data they're getting by exaggerating the perceived impact, despite the fact that it appears the game in action isn't supporting what they're suggesting.

"Notice how the frame-rate inconsistencies manifest during gameplay. This issue is difficult to observe in a 30fps video but quite noticeable during actual gameplay"

They tell us that we should notice how frame-rate inconsistencies manifest during gameplay based on the video they've provided, and then immediately afterwards tell us how difficult to observe the issue is in a 30fps video. So, which is it? The video either proves on some level what they're saying to be true, or it doesn't. Could it be the only thing from that video that even remotely comes close to proving any of the claims they're making is that framerate number that they themselves provide in the upper left hand part of the screen? Forgive me, but it sounds like they've thought up some bs to defend the integrity of their analysis. If it's truly as noticeable as they say, why not go the extra mile and provide us a proper 60fps video that would prove how very noticeable this is? Because, really, with the exception of their own number that they provide in the upper left portion of this video, nothing in the video supports their claim as far as I can see.

I've always believed to a degree that Digital Foundry's framerate tests sometimes felt as if they were questionable or outright flawed in terms of the data they gave us, and how they then claimed that data directly translates to the actual user experience, and I think this article strongly supports that belief. I may very well notice what they claim when I see this game for myself - and I will considering I bought a copy for my sister in law, which is visiting me for a week starting June - but until I see that for myself, I'm calling bogus on their claims as to how noticeable this supposedly is during actual gameplay.

What the...even just knowing what they are describing I already know how it looks from experience. It you are sensible to fraemrate issues, you will notice it. You can clearly see it in the WebFMs posted here.
 

EatChildren

Currently polling second in Australia's federal election (first in the Gold Coast), this feral may one day be your Bogan King.
Here is a terrifically bad image made by yours truly to visually illustrate what Digital Foundry appear to be describing.

mk806sd5.png


The top image is locked 60fps. Red/blue indicating new frames. In this instance we see sixty new frames are rendered in consistent sequence for each of the two seconds.
Middle image is locked 30fps. Red/blue again indicating new frames, same consistency, but as you can see there's only half the data by comparison.
It doesn't matter if the game is a different number to these. It could be 40fps, 50fps, 59fps, 43fps, whatever. The frame count fits into each second.

The last image is Mario Kart 8. Based on Digital Foundry's observation the 64th frame is a repeat of the previous frame. This is not technically 59fps. 59fps would be like the top two images: new alternating frames throughout the time period. This is not true for Mario Kart 8. Look at the start of the second bar, and you can see two red frames next to each other. That's what's happening. The scene is refreshing sixty times per second each time, but the 64th frame in each sequence is, for some reason, a double up. Not just a drop in frame rate, but a repeated frame.

This is what we call stutter/judder and can be quite annoying for some people. It's also unusual for a game that seems to otherwise render a full 60 individual frames in a second. So yeah, I guess it's a technical bug.

So is it an issue with the framebuffer? That should be fixable.

Seems to be, and yes it should be fixable.
 

Griss

Member
The reaction to DF articles is crazy embarrassing on here almost without exception. They must have thick skins over there, every single video game fanboy in the world hates them and thinks they're biased against their company of choice.

What I took from the article:
-It looks great and runs great
-The lack of AA is admittedly disappointing and the biggest issue
-It uses a clever solution for 4 player splitscreen
-There is currently a repeating frame every 64 frames that may or may not be an issue to an end user and will hopefully be patched. This falls a bit short of Nintendo's typical technical genius and so is worth remarking on for that alone.
 

dark10x

Digital Foundry pixel pusher
We appear to have a situation where their tools or methods of analysis are telling them one thing, and then they're desperately trying to seek ways to rationalize the data they're getting by exaggerating the perceived impact, despite the fact that it appears the game in action isn't supporting what they're suggesting.
This is completely, utterly, 100% incorrect.

I noticed this flaw immediately. I didn't require any tools to inform me that there was a problem. It was obvious and I found it distracting.

When I ran it through the tool set it simply made visible in a graph what I was seeing with my own eyes.

So why is it difficult to observe in Youtube videos? Two reasons, really. For one thing, Youtube playback isn't great for observing frame-rate issues as additional issues tend to crop up due to problems with the player itself. Youtube videos regularly show frame skipping and stuttering during playback that may not have existed in the original video. Secondly, Youtube cuts the frame-rate in half which means each frame should (if the player was working properly) persist for 33ms. The problem with Mario Kart is that a single frame persists for 33ms following a run of frames that persist for just 16.7ms. Youtube cannot display content at this speed and, as a result, everything is reduced to 33ms (again, theoretically as Youtube really just don't work great as a video player but it can still show the relevant data).
 
Aside from this stutter, everyone keeps saying this game is gorgeous. But it's 720p without AA, what magic are they using there at nintendo? I guess I'll have to see it myself next week.
 

DECK'ARD

The Amiga Brotherhood
Seems to be, and yes it should be fixable.

With it being specifically related to CPU players it's more likely they are using the 64th frame to update the AI rather than everything else.

So not a bug to be fixed, more a cheat to ensure the game runs at effectively full frame rate.

The AI is being updated 1/64 frames rather than every frame, probably because it was taking too long and causing a drop to 30fps.
 

JordanN

Banned
Digital Foundry said:
We can finally put that rumour to rest right here and confirm that Mario Kart 8 instead operates at what is effectively the console's standard 1280x720. Of course, considering the quality of the visuals, this can hardly be considered a disappointment especially when other developers are struggling to hit 1080p consistently on more powerful hardware.
What a shitty defense.

"Who cares the Wii did 480p, because certain PS3/360 titles had frame rate drops, despite looking massively better as a whole". I accept the Wii U for what it is but please don't embellish it.

I also thought this was funny.
Digital Foundry said:
i6LrJpHcCFDym.jpg

7/9 Here we observe the way light interacts with materials. Notice the sunlight glinting off Mario's kart. Also note the clean shadow, soft particles, and tyre tracks.
It looks like bloom/emissive material/specular power. Which can be done on PS3/360 and even Vita.
It's not even the good kind either.
 

The Boat

Member
Here is a terrifically bad image made by yours truly to visually illustrate what Digital Foundry appear to be describing.


The top image is locked 60fps. Red/blue indicating new frames. In this instance we see sixty new frames are rendered in consistent sequence for each of the two seconds.
Middle image is locked 30fps. Red/blue again indicating new frames, same consistency, but as you can see there's only half the data by comparison.
It doesn't matter if the game is a different number to these. It could be 40fps, 50fps, 59fps, 43fps, whatever. The frame count fits into each second.

The last image is Mario Kart 8. Based on Digital Foundry's observation the 64th frame is a repeat of the previous frame. This is not technically 59fps. 59fps would be like the top two images: new alternating frames throughout the time period. This is not true for Mario Kart 8. Look at the start of the second bar, and you can see two red frames next to each other. That's what's happening. The scene is refreshing sixty times per second each time, but the 64th frame in each sequence is, for some reason, a double up. Not just a drop in frame rate, but a repeated frame.

This is what we call stutter/judder and can be quite annoying for some people. It's also unusual for a game that seems to otherwise render a full 60 individual frames in a second. So yeah, I guess it's a technical bug.



Seems to be, and yes it should be fixable.
Thank you for this.
Aside from this stutter, everyone keeps saying this game is gorgeous. But it's 720p without AA, what magic are they using there at nintendo? I guess I'll have to see it myself next week.
There's no magic here. A game with a good art direction that shows technical proficiency can look good despite some flaws. 720p doesn't instantly render a game ugly and in this case, the lack of AA is barely noticeable in motion for most people, just as very few people notice this judder.
 

10k

Banned
720p59. So close Nintendo, so close.

My tv refreshes at 60 so I may notice the jarring because it's not synced at a multiple of 30. We will see.
 
What a shitty defense.

"Who cares the Wii did 480p, because certain PS3/360 titles had frame rate drops, despite looking massively better as a whole". I accept the Wii U for what it is but please don't embellish it.

I also thought this was funny.

It looks like bloom/emissive material/specular power. Which can be done on PS3/360 and even Vita.
It's not even the good kind either.

I take it you are serious on this?
 
Top Bottom