Digital Foundry Article Technical Discussion Archive [2015]

[Deleted member 11852]

IIRC, the cut-scenes still showed a higher framerate for PS4 (still a GPU advantage there I think?).

I meant unexpected but wasn't aware of any cutscene differences.

 

[function]

Trying to match the PS4 resolution on Xbox One wasn't such a great idea in this case.

Knock 20% off the resolution and the game would probably have run fine.

 

[Bain_Kaos]

Trying to match the PS4 resolution on Xbox One wasn't such a great idea in this case.

Knock 20% off the resolution and the game would probably have run fine.

So, like Watch_Dogs? (900p vs 792p)

 

[Shortbread]

Trying to match the PS4 resolution on Xbox One wasn't such a great idea in this case.

Knock 20% off the resolution and the game would probably have run fine.

Maybe both using a dynamic resolution solution? Scale to the highest possible resolution during nonstressful scenes, and rubber band back where needed during more stressful events.

 

[Recop]

Maybe both using a dynamic resolution solution? Scale to the highest possible resolution during nonstressful scenes, and rubber band back where needed during more stressful events.

Personally, i prefer a static resolution.

Dynamic resolution is like chromatic aberration : not everyone taste.

 

[pjbliverpool]

Maybe DF's comparisons aren't very accurate ? They might make mistakes on true console settings ?

They post comparison screenshots and video's to support their articles so that seems like a stretch. They'd also have to be very wrong in every single face off they post since they pretty consistently equate the PS4's performance to <7870 performance (exactly where you'd expect it to be) which is much slower than the 280x.

This is just one example (although many, many others could be posted) but take Battlefront which according to the recent face off runs at the equivalent of 'High' settings on the PS4. DF's comparison screenshots seem to support that quite thoroughly. At high settings and 1080p (which is higher than the PS4's 900p) a R9 285 (slower than a 280x) averages 80fps. So to expect the PS4 to perform at or above the level of a 280x in this games seems unrealistic to say the least. Yes this is just one example but virtually all other games paint the same picture.

http://www.gamersnexus.net/game-bench/2130-star-wars-battlefront-pc-fps-benchmark-graphics-cards

 

[psorcerer]

Newest fun article
http://www.eurogamer.net/articles/digitalfoundry-2015-vs-texture-filtering-on-console

TL;DR we couldn't prove by any means that console hardware has any problem with anisotropic filtering, but still CONSOLE HARDWARE IS A CULPRIT, PC MASTER RACE FTW!!1111

 

[Shortbread]

Newest fun article
http://www.eurogamer.net/articles/digitalfoundry-2015-vs-texture-filtering-on-console

TL;DR we couldn't prove by any means that console hardware has any problem with anisotropic filtering, but still CONSOLE HARDWARE IS A CULPRIT, PC MASTER RACE FTW!!1111

Moved to another thread...

The Great PS4 Missing AF Mystery *spawn

 

[function]

The event horizon of the aniso-hole is continuing to goatse.

 

[Recop]

They post comparison screenshots and video's to support their articles so that seems like a stretch. They'd also have to be very wrong in every single face off they post since they pretty consistently equate the PS4's performance to <7870 performance (exactly where you'd expect it to be) which is much slower than the 280x.

This is just one example (although many, many others could be posted) but take Battlefront which according to the recent face off runs at the equivalent of 'High' settings on the PS4. DF's comparison screenshots seem to support that quite thoroughly. At high settings and 1080p (which is higher than the PS4's 900p) a R9 285 (slower than a 280x) averages 80fps. So to expect the PS4 to perform at or above the level of a 280x in this games seems unrealistic to say the least. Yes this is just one example but virtually all other games paint the same picture.

http://www.gamersnexus.net/game-bench/2130-star-wars-battlefront-pc-fps-benchmark-graphics-cards

Indeed.

 

[Deleted member 11852]

What works and what doesn't in Halo 5: Guardians
Digital Foundry presents its final analysis of the Master Chief's full Xbox One debut.

 

[TheWretched]

Just watched the 2 DF videos on Youtube. Gotta say, still looks good, even with the restrictions placed on the developer by 60Hz refresh. But... on the other hand, the VERY low LOD bias, the VERY low shadow distance and resolution, as well as the VERY low textures in places... it's really sad that it's so inconsistent. Would love to play it on PC... but even with MS touting more PC love (which they do on an annual basis), it isn't going to happen.

The dynamic resolution isn't as noticeable, at least with "Youtube AA".

 

[AlNom]

Not particularly enthused about the 60fps mantra when the animations can drop to half-rate so easily.

 

[Scott_Arm]

Really curious to see gameplay footage with maybe an overlay that shows what resolution it's running at that particular moment.

 

[HTupolev]

Really curious to see gameplay footage with maybe an overlay that shows what resolution it's running at that particular moment.

Even cooler than the results would be the implication that someone had managed to put together an automated pixel counter.

 

[Scott_Arm]

Even cooler than the results would be the implication that someone had managed to put together an automated pixel counter.

Well, yah, there's that problem. Aslo reading about the game, it sounds like they can scale to a lot of different resolutions, so it sounds like it would be impossible. It's not the case of 2 or 3 resolutions that are possible.

 

[HTupolev]

Aslo reading about the game, it sounds like they can scale to a lot of different resolutions, so it sounds like it would be impossible. It's not the case of 2 or 3 resolutions that are possible.

If you're thinking it would be an issue of representation, that could be managed. Showing the resolutions of the two axes separately would be fairly readable, or you could just multiply them and show the total pixel count to reduce things to one number.

The only real hurdle is the difficulty of the pixel counting. 343i could make a resolution analysis video quite easily if they wanted to.

 

[Scott_Arm]

If you're thinking it would be an issue of representation, that could be managed. Showing the resolutions of the two axes separately would be fairly readable, or you could just multiply them and show the total pixel count to reduce things to one number.

The only real hurdle is the difficulty of the pixel counting. 343i could make a resolution analysis video quite easily if they wanted to.

I just meant in terms of pixel counting. Too many options with too frequent changes and it's basically impossible. Still not totally sure how they do it, or why they'd scale different axes at different times. I thought maybe they just had hints for certain areas on maps that scaled the res because they know there are a large number of enemies and the possibility of a lot of effects etc, or maybe certain views or whatnot. Sounds more like the engine is doing something to determine the best resolution on the fly and scaling accordingly.

 

[AlNom]

It'd probably be far more flexible for development if it were just a performance metric rather than a locale-based trigger.

Independent axis scaling is probably so the overall upscale isn't too skewed e.g. 960x1080 (pixel doubling in horizontal)

 

[Globalisateur]

Not particularly enthused about the 60fps mantra when the animations can drop to half-rate so easily.

1080pr @ 60fpr