146 replies to this topic

[aniviron]

Hey Russ, one of the first questions asked during the town hall meeting was about the decline in visual quality from beta to present; I wasn't the one who asked the question, but it's something that is on my mind as well, so I hope I can add some things. As a preface, I have a pretty decent rig, and run on maxed everything, so I'm not cutting corners.

EDIT: The question was asked by SmokinDave73. Here's his reddit thread on the matter!

One of my personal biggest disappointments is the removal of inverse kinematics for mechs' legs. For those who don't know what this is: when walking on uneven terrain, IK bends the feet and knees so that they match the slope of the surface and neither stick through or float above it. For example, a screenshot I took of what MWO looks like today:



As you can see, one leg is floating above the mountain, while the other is embedded in it. The legs are posed as though the mech was on flat ground.

Compare that to the following screenshot, where the front right foot of the catapult is aligned to the ground, and the knee is deeply bent so that the back foot doesn't float above the ground or the front foot isn't sticking into it.



Here's another shot of IK at work from the official media on the MWO site; note how the Dragon's left knee is bent so that while the right leg is fully extended the mech stays level without floating or being embedded in the ground.

Another thing to note is that the damage textures on the Catapult's side torso and hip show off the metal underneath the armor and paint; while they weren't terribly high-res textures, they were a glimpse of the mech's internals after the outside had been peeled off. At some point, those textures were changed to a generic bullet impact texture that is a flat grey and shows as bullet holes regardless of whether the damage was caused by a gauss rifle, a flamer, or a laser.

In regards to causing that damage, internal ammo explosions have lost their 'pop.' Here's a video of an ammo explosion from the outside in beta (starts at 20s)

http://youtu.be/rgJ2n5sEn9c?t=20s

and another of the explosion from the inside (starts at 5:20):

http://youtu.be/syyrgIKLim8?t=5m20s

Regarding the ammo itself, the missile flight path changes have been pretty disappointing from a visual standpoint. The swarm effect, the corkscrew, and the undulating paths that we've had at different points in the game all lent missile fire a good amount of visual interest (and gave SRMs a nice added bit of skill ceiling, while leaving the floor intact). The current straight-line flight paths are usable, but ultimately severely lacking in visual interest.

In regards to mapping, it's a bit of a mixed bag. Some pretty cool things were removed, like the blizzard rolling in on Frozen City; now the weather effect just starts and stops, but at one point, you could watch it rolling in over the frozen sea before it hit. Now it's just a fog and snow effect that fades in an out during the match.

After some LOD passes were done, Caustic and River City both looked heinously bad, with buildings and rocks popping in and out of existence on a whim. To PGI's credit, it's been fixed on River City, and the map now looks better than it did before, at least in regards to building LOD. Caustic however, is still awful. Walking in a straight line yields the ground and rocks undulating underfoot every hundred meters or so, changing size and shape as distance closes.

Even Alpine, added quite a while after the old cadre of maps, has had some tuning: the rock texture on the sides of the mountains used to have a depth mapping effect to make the map look less like a series of gentle rolling hills with white painted on the grass.

Another thing that's missing is map interactivity: while it was never huge, there were some great little details that really made them feel a lot better. Watch the following video, starting at 29s:

http://youtu.be/zWZBLA-Y8zg?t=34s

What you're looking for isn't the collision on the mechs; it's under them. The ground used to crater and crack when mechs fell over. And surprise of all surprises, it even used to knock streetlights down if you fell over on them!

Go ahead and rewatch that video and tell me if something feels off about River City to you while you're at it, because it should: RC is gorgeous in that video. It's the same map, more or less, but that wonderful feeling of light and reflection and depth and contrast that the whole video has is something the RC we have now lacks. Well, I thought to myself, maybe it was Kon's settings? Or the video encode? Maybe my memory is just getting dull?

So I went hunting. Here's another screenshot from the official media section of the MWO website. It's got that same feeling as Koniving's video; look at how the glass reflects the light, the way it shines off the side of that catapult, and the great contrast, presumably from the HDR implementation.



Now here's a screenshot of that same intersection (sans mechs) that I took ten minutes ago.



Compared to the first image, it looks... flat. Drab. Shabby. Uninteresting. The windows hardly reflect, and what they do looks bad. The soft fog is gone, the colour palette is less vibrant. It looks like a knock-off of the game the first image is from. It doesn't look bad, per se, but it's not gorgeous like that first shot.

Look at the lighting on the mechs in the first image too- see the way that the Hunchback's side is lit, but the back is dim? Compare that to the first screenshot in the post, and notice how the lighting effect is almost completely gone. The Hunchback on Alpine is lit in a flat, even manner, with every surface as bright as every other. Look back at the beta screenshot again, and the shadowing and bright spots on the shutdown founder's Cat. That kind of light and shadow detail seems to be lostech.

It probably helps in those screenshots that back then, every mech got a custom texture that really drew out the beauty of the model- the stock green camo that comes on most mechs now has nowhere near the same amount of finesse. Furthermore, the older textures seem to have been higher-res; what we have now is a crapshoot, with sizes ranging from 512x512 to 2048x2048. Here's a great post by Lordred showing the various texture sizes on all the mechs.

So Russ, when that question was asked, I have to assume this is what the asker was referencing (and feel free to correct me if I'm wrong). It just feels like the visual quality has slipped significantly since beta in all sorts of little ways; the mech animations don't have the same detail they used to, ER Mlasers cast a green light despite burning yellow, snow shows as white in thermal as though it were hot, etc. It's telling that in a recent video review, TotalBiscuit said of a game that it looked great, like every CryEngine game, and then went on to add, "except Mechwarrior Online."

Addendum: I was going to add to the previous paragraph that mechs no longer left individual footprint decals like they used to, but now they just don't leave footprints at all. Add one more thing to the list of details that have been erased.

Edited by aniviron, 06 October 2014 - 09:09 AM.

[Alistair Winter]

Good post. Images not working though.

[aniviron]

Alistair Winter, on 03 October 2014 - 02:19 AM, said:

Good post. Images not working though.

Switched image hosts. Any better?

[Vassago Rain]

Most of it's due to the 'dual core optimizations,' which needed further optimizations to make the game playable again.

Also, in february 2013, they overhauled the mech skins to fit with the new camo system. That's why the old damage meshes are no longer in existence.

2014.


2012.


Perceptive viewers will note that the atlas damage mesh of old also showed of the diamond armor patterns underneath the paint. Especially notable on the head, just above the atlas' right eye.

Edit: here's a better picture of the oldtimey kinematics.



You'll also note that at one point, you could see stuff on forest.

Edited by Vassago Rain, 03 October 2014 - 02:31 AM.

[Crunk Prime]

I think you're on a bit of an island there bud, these generic low res bullet hole textures put the mechs in the best state they've ever been!

Edited by Crunk Prime, 03 October 2014 - 02:42 AM.

[aniviron]

Vassago Rain, on 03 October 2014 - 02:28 AM, said:

Most of it's due to the 'dual core optimizations,' which needed further optimizations to make the game playable again.

Also, in february 2013, they overhauled the mech skins to fit with the new camo system. That's why the old damage meshes are no longer in existence.

Perceptive viewers will note that the atlas damage mesh of old also showed of the diamond armor patterns underneath the paint. Especially notable on the head, just above the atlas' right eye.

Edit: here's a better picture of the oldtimey kinematics.

Yep, you're right about the cause. I realize my OP comes off as a little harsh, but I know that every change was done for good reasons, to try and optimize performance, or make way for a new feature, something like that. I'm not saying that this is a willful degradation of the game's visual fidelity, or that it happened through negligence.

It's just that every time I hear that something is being optimized in the patch notes, I dread the change. Every small level of detail removed detracts from the game experience. You might not notice it in the patch it happens, maybe at some point you'll go, "Huh, I thought that used to look better?" But all the little details add up to make a huge difference, and a lot of what made this game so attractive at one point got lost along the way.

Thanks for the pics of the kinematics, and armor damage- they're just what I was looking for. And the second one brings up something I was thinking about, but didn't add in the OP, but it fits here.

The Atlas' glowing eyes epitomize what happened to the visual fidelity of the game. They were added early on in beta. Why? Because they look badass. It let you know death was plodding towards you- time to run. Don't mess with the Atlas. They were removed later, for gameplay reasons, because they let you know that death was plodding towards you, and sometimes death wants to set up a nice ambush. It's a good reason to remove it, it makes sense, and I completely respect why the decision to remove the glow was made.

But that decision had a tradeoff. The glow-eyed Atlas was MWO's icon, it was amazing, and the game definitely lost something by removing it. The only workarounds would involve a lot of extra coding time that PGI's small team can't spare. A big studio would have maybe linked the glow to the cockpit lights which pilots can turn on and off (which is still in the game and working thank god, unlike heatsink monitors) but it would have taken time and work better spent on other things.

Tradeoffs happen. But Russ wants to know why we think the game looks worse: well, this is why.

[Vassago Rain]

aniviron, on 03 October 2014 - 02:42 AM, said:

Yep, you're right about the cause. I realize my OP comes off as a little harsh, but I know that every change was done for good reasons, to try and optimize performance, or make way for a new feature, something like that. I'm not saying that this is a willful degradation of the game's visual fidelity, or that it happened through negligence.

It's just that every time I hear that something is being optimized in the patch notes, I dread the change. Every small level of detail removed detracts from the game experience. You might not notice it in the patch it happens, maybe at some point you'll go, "Huh, I thought that used to look better?" But all the little details add up to make a huge difference, and a lot of what made this game so attractive at one point got lost along the way.

Thanks for the pics of the kinematics, and armor damage- they're just what I was looking for. And the second one brings up something I was thinking about, but didn't add in the OP, but it fits here.

The Atlas' glowing eyes epitomize what happened to the visual fidelity of the game. They were added early on in beta. Why? Because they look badass. It let you know death was plodding towards you- time to run. Don't mess with the Atlas. They were removed later, for gameplay reasons, because they let you know that death was plodding towards you, and sometimes death wants to set up a nice ambush. It's a good reason to remove it, it makes sense, and I completely respect why the decision to remove the glow was made.

But that decision had a tradeoff. The glow-eyed Atlas was MWO's icon, it was amazing, and the game definitely lost something by removing it. The only workarounds would involve a lot of extra coding time that PGI's small team can't spare. A big studio would have maybe linked the glow to the cockpit lights which pilots can turn on and off (which is still in the game and working thank god, unlike heatsink monitors) but it would have taken time and work better spent on other things.

Tradeoffs happen. But Russ wants to know why we think the game looks worse: well, this is why.

It wasn't for good reasons, though, because dual cores still can't play the game properly, and when they were doing those 'optimizations,' people had advanced to i5s and i7s. MWO is heavily CPU dependent, so your other hardware doesn't really matter.

Also it makes people question why they went cryengine in the first place. It's a super beefy engine for making the (at the time) best looking games, with similarily high requirements. Why use the engine if you're gonna take out all the pretty things? Just roll unreal like hawken.

I have friends who used to get some 30 FPS back in the olden days, but they're capped out at 20 in today's MWO, which looks a lot worse than what we used to have.

Here's my favorite comparison picture to MWO of 2014. While yes, this is crysis 3, which was made by crytek themselves, think about what you're seeing here.



In MWO, there'd be a not very subtle filter applied to everything, probably some brownish haze, the clearly defined, white armored soldiers in front of the orange supply pyramid would be black blobs thanks to LOD, the lighting from the sun would have created awful green cockpit glass effects, and you'd likely not be able to read the text on my gun.

It doesn't have to be that way, and it wasn't always that way. While I don't have access to my old closed beta screenshots, know that it wasn't always this way.

I, too, dread 'optimizations,' because it's become PGI's way to say 'we're cutting out more features.' There's no good reason for it in 2014.

Here's some more of my favorites, from back when they said picture-in-picture is an engine limitation. If you look into my gun's scope, well...





Obviously, PIP is an engineering term. What we were asking for was 'that trick they do in crysis, crysis warhead, crysis wars, crysis 2, crysis 3, and living legends that makes it seem as if you have a PIP.'

Edit: remember what they sold you on!



I'm still sore.

Edited by Vassago Rain, 03 October 2014 - 03:18 AM.

[Logan Hawke]

I am definitely interested to see the Dev's response to this. I don't think it should be a priority by any means, but I do think they should keep it in mind as something that could use work.

[Kiiyor]

Wasn't inverse kinematics canned because the servers couldn't handle all the extra calls being made for 12vs12?

[Vassago Rain]

Kiiyor, on 03 October 2014 - 03:30 AM, said:

Wasn't inverse kinematics canned because the servers couldn't handle all the extra calls being made for 12vs12?

It was killed long before 12v12.

[Jonny Taco]

Just listened to Russ's answer about this general question on the town hall tiwtch stream and I must say... If Russ really thinks that the new damage textures look better than the old, He's the one on the island here.

[Kiiyor]

Vassago Rain, on 03 October 2014 - 03:32 AM, said:

It was killed long before 12v12.

I swear I can remember the devs talking about it, but all I can find is this (moderately related)

Matthew Craig, on 11 September 2014 - 10:08 AM, said:

First off I'd like to thank Bishop Steiner for the tone of this thread I think it is very fair and we've never said we are perfect as devs, we're human like anyone else. This is the first thread I've seen in a while that has encouraged me to respond so again thanks for that Bishop Steiner.

The reason for my response is to help clarify some of the speculation around engine choices. Firstly allow me to clarify Russ' statement about the engine being CPU bound, it is true that our game is currently CPU bound, Is it true that it has to be that way because of the engine? no not really but there is truth in the detail allow me to explain.

One of Mechwarriors challenges is that being an FPS it is regularly compared to other FPS titles e.g. if Counterstrike can get hit detection right then why can't MWO. There are a couple of notable differences with MWO the first being that each Mech can field a wide variety of weapons, in a game like CS you don't fire 6 lasers and 2 LRM banks all at the same time with AMS going off etc.

This adds to the complexity of the simulation that the server is doing and the overall workload for host state rewind and networking this is partly what limits us to 12 vs. 12 currently and made it such a challenge in the first place.

Also Mechs are not constructed in single parts and don't behave as simple characters this meant that earlier in the project when Mechs were being setup the approach that was taken was to compose them of separate attachments to give us flexibility to animate separate components, blow them off etc. over time we realized that this isn't ideal for optimal draw call use.

Most modern engines CE3 and UE4 included use deferred renderers. This means that there are multiple passes that objects get rendered in, so any object using a notable number of draw calls gets amplified e.g. it has to be rendered in the ZPass, Shadow Pass, General Pass and potentially Post Processing Pass which can amplify the draw call usage.

Draw calls remain very expensive on the CPU side and much of the research and development currently going into Mantle and DX12 is aimed at reducing overhead for draw call usage (and API / driver overhead in general) so that games can use more draw calls and focus on creating great consumer experiences without having to get overly concerned with hitting draw call barriers.

So would another engine have had similar problems to CE3? potentially as mentioned the character editor initially controlled how we composed the Mechs and setting them up optimally for both performance and flexibility were opposing goals in CE3 where another engine may allow an easier path to attain both goals.

With respect to why CE3 or UE4 there are many factors at play, familiarity with the engine not being as large a factor as many would think as most modern engines are very similar in terms of feature sets and tools they provide though that isn't to say there aren't notable differences.

Everyone that cares to follow engine development knows that the last few years has seen major tectonic shifts in that area not only in terms of technology but also in terms of licensing and like any game studio we actively follow developments and respond accordingly. The new project has very different requirements from MWO and we evaluated from the available engines what was the best fit for that project and we have been very impressed as have many with the tool set that Epic has put together for UE4.

Does that mean we would want to do things differently for MWO, well the premise is unsound, in that the landscape as mentioned was very different when we evaluated tools for MWO and at this point in MWO development switching engines is highly unlikely for obvious reasons. As many of you can appreciate we have heavily modified CryEngine at this point to suit the needs of MWO and so it is hardly a stock version any more, also our goal is as it has been for a long time to attain a very stable and functional environment for all the players playing MWO on a daily basis; long gone are the days of Closed Beta and we appreciate that and aim every day and every release to make the game more stable and reliable.

We've only done one major engine update since Closed Beta and that was to ensure we got working DX11 support which we still hope to leverage further in the years to come, as Russ mentioned in the town hall so long as there are MWO fans that want to play we will continue to aim to improve and expand MWO.

We know we've had some major struggles in this respect and many have involved fighting with the engine technology e.g. the long standing Scaleform issue we had (disco polys); however, that can't be entirely laid at the engines feet as any new game exercises an engine in potentially new ways that the engine developer can't foresee. We feel that hopefully once we get on top of the current de sync issues bar any further issues that arise from CW we will be reaching a point of maturity with the technology for MWO where we can begin to refine the base of MWO e.g. further UI 2.0 improvements both in the form of fixing bugs and functionality.

We also know that there are many still running older hardware that would like MWO to run more efficiently than it currently does and truth is that we could likely do more with more art resources dedicated to assisting optimizing existing assets. The general request from the community though is for new assets (maps/mechs) which shifts lots of resources to new assets and doesn't allow for there to be much internal pressure to improve older assets.

Something worth noting that I would like to correct is the notion that performance for the game has degraded over time and that performance was much better back in Closed Beta. We gather telemetry on many aspects of the game including average user fps and the truth is that while there have been both jumps forwards and backwards in terms of performance the trend line is net positive since Closed Beta.

It is likely the odd regressions and the fact that it can be true for a small number of users that their particular performance may be lower than earlier in the project that leads to the conclusion that it must be that way for everyone when overall that isn't the case.

We will continue to do our best to ensure the game is stable and runs great and if you ever ask yourself if we (the devs) are aware of a particular issue or care the answer is overwhelmingly yes we do care and are painfully aware, if something takes a long time to get resolved it is generally either because it is proving to be very difficult to fix or there are other competing priorities (many of which the community don't get visibility on e.g. improving customer support tools, dealing with billing issues etc. etc.) that need fixing first.

Hopefully that helps to give some more detail to the debate.

I bolded the part that I think is the most relevant. I recall now; it was all part of trying to improve hit detection and HSR. I think.

[meteorol]

Well, i was playing with 9 FPS during CB, so i can somewhat understand they toned down the graphics.
In my eyes, MWO was always much more taxing to the system than it should have been, considering the graphical state at the time, and especially much more taxing than any other game on the f2p market.
When i ran it with 9FPS, it didn't offer graphics that would justify 9FPS. Even now after they toned down the graphics, it runs at lower FPS than you would expect given how it looks. It almost seems they had the choice between worse graphic or better optimization and did the first, because the latter would have taken too much time/money/was outside of their abilities.

Edited by meteorol, 03 October 2014 - 03:59 AM.

[RedDragon]

Vassago Rain, on 03 October 2014 - 03:13 AM, said:

Obviously, PIP is an engineering term. What we were asking for was 'that trick they do in crysis, crysis warhead, crysis wars, crysis 2, crysis 3, and living legends that makes it seem as if you have a PIP.'

To be fair, that on the image looks like it's just a copy of the already rendered scene on your screen. The problem with PIP, as I understand, is that it is not possible to render a different view on the main screen. And that is something that would be needed for proper zoom, and that's also why current zoom is just an enlargement of the things under your crosshairs, not a separately rendered zoom window as it should be. Why the engine doesn't support this in the first place is completely beyond me, zoom and additional screens are part of FPSs for decades now.
Still, not having PIP is no reason why we for example don't have a rear view yet. Doesn't have to be PIP, as MW4 has shown.

Another part of the problem with performance is 12v12 I guess. I used to get good frame rates back in Beta, but 8 more mechs on the field definitely has an impact on performance.
But again - this is no real excuse for why the game looks so much crappier than in Beta and on the other hand runs like sh*t on most rigs.

[Vassago Rain]

RedDragon, on 03 October 2014 - 03:58 AM, said:

To be fair, that on the image looks like it's just a copy of the already rendered scene on your screen. The problem with PIP, as I understand, is that it is not possible to render a different view on the main screen. And that is something that would be needed for proper zoom, and that's also why current zoom is just an enlargement of the things under your crosshairs, not a separately rendered zoom window as it should be. Why the engine doesn't support this in the first place is completely beyond me, zoom and additional screens are part of FPSs for decades now.
Still, not having PIP is no reason why we for example don't have a rear view yet. Doesn't have to be PIP, as MW4 has shown.

Another part of the problem with performance is 12v12 I guess. I used to get good frame rates back in Beta, but 8 more mechs on the field definitely has an impact on performance.
But again - this is no real excuse for why the game looks so much crappier than in Beta and on the other hand runs like sh*t on most rigs.

It's a smoke and mirrors effect, but it does what you'd expect, and what people were asking for.

The point is they could have easily copied what's been used in every cryengine game since forever, rather than insist on keeping the 8-bit advanced zoom around unchanged for so long, while citing engine limitations.

Maybe it doesn't support this technology, but why can every other game in the same engine have something that LOOKS like what's being asked for? It's why I said PIP is an engineering term.

[RedDragon]

Vassago Rain, on 03 October 2014 - 04:04 AM, said:

It's a smoke and mirrors effect, but it does what you'd expect, and what people were asking for.

The point is they could have easily copied what's been used in every cryengine game since forever, rather than insist on keeping the 8-bit advanced zoom around unchanged for so long, while citing engine limitations.

Maybe it doesn't support this technology, but why can every other game in the same engine have something that LOOKS like what's being asked for? It's why I said PIP is an engineering term.

Erm... if a zoom doesn't really magnify the picture, what is the point? People were asking for a real zoom, not a fake one.

[EvilCow]

I understand toning down the graphic because performance concerns, but why do that at all quality levels? high level quality settings could have retained all the advanced effects.

MWO looks quite outdated already and this is bad for a game supposed to last some years.

Edited by EvilCow, 03 October 2014 - 04:16 AM.

[Ghogiel]

Ironically when the OP noted the window reflections in river city, the original implimentation without environment probes was rubbish. I noted this back in closed beta. And was happy that something was done about it. Though the lighting overhaul (possibly not only done due to aesthetic but also odd things and changes going on in the CE3 versions at the time, might have warranted the overhaul) is debatably worse looking.

[Colby Boucher]

So, Russ- you wanted an explanation of how the game looked better in closed beta? Here you go! Could we get a better answer now?

Edit: Seriously, anyone who's reading this, post something, just keep this thread alive so Russ notices and it doesn't get buried!

Edited by Colby Boucher, 03 October 2014 - 04:33 AM.

[DEN_Ninja]

Colby Boucher, on 03 October 2014 - 04:30 AM, said:

So, Russ- you wanted an explanation of how the game looked better in closed beta? Here you go! Could we get a better answer now?

It is like 5:30am in vancouver so...check back in like 3-5 hours.