Battlefield V reflection tech *spawn

[Entropy]

Seriously, try to play a game with no or little shadows and see how distracting and visually wrong that is, your depth perception will be completely disrupted.

Of course shadows help ”anchor” the objects in the 3D scene. I was just claiming that the minutiae ultimately just isn’t all that important for us mammals because we are simply more interested in the objects themselves rather than the shadows they cast, and I’d guess that’s true for any creatures.

I won’t belabour the point.

 

[Shifty Geezer]

Obviously you need to see the reflections in action with actual hands on to properly judge it, seeing it from the eyes of a compressed playback is not ideal at all.

Quite the opposite. If your intention is an objective tech analysis, you need to not play it, where you'll be focussed in particular areas, and review recorded footage allowing you to correct misconceptions (like me first impression for full framerate video that the reflections weren't being updated in realtime). That's what DF et al do.

Whether I was playing the game or not, all these artefacts would be just as distracting. The important point here is pure raytracing won't have these issues, so it shows how the current limited attainable performance and traditional real-time workarounds, barring future optimisations, are far from the ideal. And before everyone says, "it's early days yet," I thought B3D was about looking at the hardware and software and analysing it. I've presented four videos showcasing issues with the rendering to factor into the performance considerations of realtime raytracing, in particular noting that this is the highest end 2080 Ti and it'd be well worth comparing to the 2070 to see how the performance differs. If that's not exactly the sort of discussion you're after, why are you on B3D??

You say you don't see the reflection on the floor. I present a video highlighting it. You don't respond. What's your interpretation of the glitches? Are these raytraced reflections going wrong, or hybrid screen-space reflections which muddies the contribution from RTX?

 

[DavidGraham]

Quite the opposite. If your intention is an objective tech analysis, you need to not play it, where you'll be focussed in particular areas, and review recorded footage allowing you to correct misconceptions (like me first impression for full framerate video that the reflections weren't being updated in realtime). That's what DF et al do.

Didn't mean for you to play it, I meant to analyze it hands on, not through a compressed footage. To have the opportunity to look at it from various different angles and to zoom in and out if necessary.

Are these raytraced reflections going wrong, or hybrid screen-space reflections which muddies the contribution from RTX?

To determine that, again you need to test it hands on.

Whether I was playing the game or not, all these artifacts would be just as distracting.

Yes, and as I was pointing out, you complained these issues are distracting, I just reminded you that without ray tracing, a metric ton more distracting artifacts would occur.

And before everyone says, "it's early days yet," I thought B3D was about looking at the hardware and software and analyzing it

And then repeat the the embarrassment that occurred when many people shouted RTX= only 1080p60?! Jumping to conclusions without actual data always leads to embarrassments.

in particular noting that this is the highest end 2080 Ti and it'd be well worth comparing to the 2070 to see how the performance differs.

We already know the performance difference between a 2070 and a 2080Ti, I assume you meant the quality difference?

 

[Shifty Geezer]

Yes, and as I was pointing out, you complained these issues are distracting, I just reminded you that without ray tracing, a metric ton more distracting artifacts would occur.

I never said it wouldn't. It's just the raytracing is going to solve these eventually. How far are we from that? Will it be possible with RTX 2080 Ti, or will that not happen for a few generations of hardware?

And then repeat the the embarrassment that occurred when many people shouted RTX= only 1080p60?! Jumping to conclusions without actual data always leads to embarrassments.

It's not an embarrassment to be wrong. Plus no conclusions have been drawn. The only thing I've suggested has a big ol' caveat over it, and is just interpreting the current state of play. "Things may change" is a given, but if we refuse to talk about stuff until it's 'finished', we won't have much to talk about. Let's not discuss RTX until 2022...

We already know the performance difference between a 2070 and a 2080Ti, I assume you meant the quality difference?

Yes, how does that performance relate to the quality of the reflections given we know the ray counts are 15% versus 40%. Are there more artefacts as a result of less rays?

 

[Scott_Arm]

They change the roughness factor at low and medium to preserve some of the quality of the obvious reflections. I imagine mostly the differences will just look like low resolution reflections, but other differences are probably not significant.

As for drawing conclusions from battlefield v, it really comes down to understanding software. Most gamers have a bigger grasp on hardware, so they don't really understand how bfv ray tracing performance can jump significantly from one patch to the next. Plenty of people on this forum drew conclusions that were unreasonable. Even with this patch we really can't come to any conclusions about more patches, drivers, sdk and tool updates, or whether bfv will be reflective of other titles that implement reflections with dxr. Battlefield v is a game with large maps, large player counts and buildings that can be destroyed. That has an influence on how quickly the bvh goes stale.

 

[milk]

Is the current build of BFV using the zbuffer mip-chain raymarching optimization to accelerate raytracing dice developers had mentioned, or is that not in yet? Because some of the weirder artifacts there looked like the kind of thing I would expect from that kind of scheme not being fully accurate.

 

[DavidGraham]

Yes, how does that performance relate to the quality of the reflections given we know the ray counts are 15% versus 40%. Are there more artefacts as a result of less rays?

If both 2070 and 2080Ti are running ultra DXR, their rays count would be the same.

 

[vipa899]

Battlefield v is a game with large maps, large player counts and buildings that can be destroyed.

Yet it looks very impressive with RT, DF was very fond of it. Heres hope consoles sometime in the future also see it, so they can enjoy next gen graphics too.

 

[Silent_Buddha]

I disagree each time I see a bad shadow maps, it hurts my eye and it happens all time.

I hope many next gen games will have raytraced shadows using compute if no RT hardware in next gen console.

Hence, why shadows are the first thing I turn off on PC if I need the performance.

They are the most distracting if not done perfectly and the least important.

Regards,
SB

 

[Shifty Geezer]

They change the roughness factor at low and medium to preserve some of the quality of the obvious reflections. I imagine mostly the differences will just look like low resolution reflections, but other differences are probably not significant.

When there's widespread reflections like water, the area of reflective surfaces is going to be comparable.

As for drawing conclusions from battlefield v, it really comes down to understanding software. Most gamers have a bigger grasp on hardware, so they don't really understand how bfv ray tracing performance can jump significantly from one patch to the next. Plenty of people on this forum drew conclusions that were unreasonable. Even with this patch we really can't come to any conclusions about more patches, drivers, sdk and tool updates, or whether bfv will be reflective of other titles that implement reflections with dxr. Battlefield v is a game with large maps, large player counts and buildings that can be destroyed. That has an influence on how quickly the bvh goes stale.

Are you on board for a banning of all RTX discussion other than praise of what it gets right until 2022 then, to give algorithms and software a chance to mature so we can get an accurate picture?

If both 2070 and 2080Ti are running ultra DXR, their rays count would be the same.

Oh, of course, silly me. Used to consoles having fixed setting. So low quality and high quality can be compared on the same card, and it's only framerate that is affected which isn't by much.

 

[milk]

If both 2070 and 2080Ti are running ultra DXR, their rays count would be the same.

They will have the same rays per frame, but at different framerates, which does end up affecting the visual quality in the end since the reflections are so reliant on temporal reconstruction and aa.
The use of temporal accumulation really does break some of our common assumptions in real time rendering.

 

[Scott_Arm]

When there's widespread reflections like water, the area of reflective surfaces is going to be comparable.

Are you on board for a banning of all RTX discussion other than praise of what it gets right until 2022 then, to give algorithms and software a chance to mature so we can get an accurate picture?

No, I'm not on board for banning discussion of RTX. It's just the reality that new hardware requires time for the software to catch up. It's way too early to draw any sweeping conclusions. Discussing what battlefield is doing in particular is still interesting.

 

[vipa899]

Its ray tracing in games, why would anyone without an agenda want that banned, its new promising tech. RTX in general are a very good range of GPUs.

 

[digitalwanderer]

Its ray tracing in games, why would anyone without an agenda want that banned, its new promising tech. RTX in general are a very good range of GPUs.

Even if it wasn't promising it definitely counts as interesting and conversation worthy, I don't see anyone banning any talk about it.

 

[Shifty Geezer]

If people want to engage in discussion about the interesting aspecs of RTX tech, why does a constant cadre of people keep objecting to those of us who point out limitations with it? I took the time to put together four video clips showcasing issues, to correct DavidGraham where he couldn't see the reflection issues etc., yet far and away the majority of posts are about dismissing these points with zero technical discussion. Not one conclusion has been drawn by me, yet everyone's harping on about jumping to conclusions. It's very boring. So much to talk about the technology, in particular denoising limitations and what causes the artefacts, and no-one wants to bother, prefering to just gush uncompromisingly.

Someone being wrong with technical predictions is no embarassment, but this thread very much is an embarassment to B3D's raison d'être. It's being made nigh impossible to have a decent discussion on this new and exciting tech.

 

[Scott_Arm]

@Shifty Geezer I don't take issue with the clips you posted. They highlight real issues or bugs.

i think it would be interesting to know whether they fully implemented their hybrid screen-space and dxr approach and how that might contribute to visual error.

 

[DavidGraham]

Not one conclusion has been drawn by me, yet everyone's harping on about jumping to conclusions.

I see four grand conclusions in your original complaint post:

These are a result of us not having enough power for 1 ray per pixel per frame.

I can't see reflections ever working well with lighting on the current RTX cards.

I'm also guessing the rays are concentrated in important areas or when there's less reflectivity, so as more reflections are present, the quality of those reflections decrease.

There just isn't the power, seeing as reflections only are already taking considerable shortcuts.

See my point? None of those points you made have any data to back them up, you don't see quality decrease when more reflections are present, you don't know for certain the artifacts are due to 1 pixel per ray, and you have already decided reflections + lighting are impossible on RTX.

why does a constant cadre of people keep objecting to those of us who point out limitations with it?

Because your complaints always jump to grand conclusions, and this time despite not having the hands on experience or analysis. In the beginning of this thread, people complained performance is low, then performance got significantly higher, then they complained IQ was affected, then it turns out IQ is in fact increased. What's more to complain about this time? oh the rays are too low, they are causing artifacts! The hardware is weak!

It's getting really tiring. People here are simply asking for an objective analysis without jumping to massively grand conclusions, all you could have done is point out the artifacts and then started theorizing about their causes, people would have chimed in and a conversation is started, but how can we start one when you've already decided the problem, the cause and the effect?

 

[vipa899]

I think its always possible to find disadvantages with any tech, in special first implementations. Studying images of the net and you will find errors, even in Sony games.
RT is all but perfect but thats to expect, it doesnt suprise anyone. Digital Foundry thinks it is very impressive with the RT enabled, in special after the patch/performance and quality improvements. 1440p 60fps with RT in a fast online 64 player online shooter isnt that bad with some artefacts/limitations. The first hardware pixel/vertex shaders were more limited too compared to software implementations.
I dont think anyones dismissing RT limitations either directly.

There will be future titles with RT support, perhaps Metro Exodus will have a more solid ironed out implementation, as its more of a single player adventure instead of MP which BF always has been about. RT is bolted on also on BFV, yet still impresses.

this thread very much is an embarassment to B3D's raison d'être

Exactly, its going overboard by proposing banning discussions about tech, even if its imperfect in any way.
We shouldnt forget either that none of us that likes new future tech arent liking it because of supriority to consoles in any way.

All of us have an agenda somehow, or prefer a tech or platform, but your having the mod status too.

 

[Deleted member 2197]

@Shifty Geezer I don't take issue with the clips you posted. They highlight real issues or bugs.
i think it would be interesting to know whether they fully implemented their hybrid screen-space and dxr approach and how that might contribute to visual error.

Agreed. It would definitely be more constructive hearing from dev's actively engaged with the technology instead of guessing what is really is happening behind the scenes. New tech means a plethora of new techniques to be learned before an optimal solution is applied and I'd venture to say engaged developers have only begun to chip at the tip of the iceberg.

 

[iroboto]

If I could interject a bit here. I think in this particular case, we need to do a hard separation between DXR/RT the hardware, from the implementation by the devs. When we're discussing RT performance right now, we're just discussing BFV RT performance. Nothing more. It is the singular and only data point we have and we have no method of comparison to other titles right now.

We don't judge rasterization as a whole by the performance of 1 single game on 3 video cards, and I know it's not the intention to do so by anyone making arguments here, but in the hastening of our typing, we may be using language that is applying to all of RTX performance, when we are talking about just BFV right now.

To criticize BFV is absolutely legitimate, we can say their implementation has artefact, some poorer performance, etc. But we can't say that DXR or RT has artefacts or poorer performance, we don't know if it's driver performance/problems or API problems, or if it sits with BFV. I'm willing to bet it's more probable to be on the developer side, so it's important to make this distinction. I think that if we are more specific we become aligned on the same topic and people shouldn't get defensive.