Witcher 3 MUST be DX11

+
PrinceofNothing said:
Microsoft no longer even supports Windows XP, so why should games support it?

Ummm speaking as a Microsoft employee. Yes we do. Until April of next year to be exact. Please make sure of your facts before touting them. Thank you.
 
Ancient76 said:
Actually you are wrong.

Am I now?

To have better shadows, lighting, texture repetition, better lod, you don't need DX11 at all.

DX11 has soft shadows by default, which can be done in DX9, but the performance cost is too great which is why developers never use it in DX9 mode.

Lighting? DX11 uses 64 bit HDR for more precise, and authentic looking lighting. DX9 is stuck with 32 bit precision, which doesn't look nearly as good.

Texture repetition? DX11 has a higher and more efficient texture compression ratio, which means they can use more detailed and varied textures without overwhelming the VRAM..

Better LOD? Never heard of tessellation I take it? Also, DX11 drastically reduces or eliminates object pop in when done correctly due to a more efficient rendering pipeline with less overhead.
 
Kyriene said:
Ummm speaking as a Microsoft employee. Yes we do. Until April of next year to be exact. Please make sure of your facts before touting them. Thank you.

Fine I was wrong, but since the Witcher 3 will likely ship in the 3rd of 4th quarter next year after official support for XP ends, then what I said still has merit.

There's no point in releasing a new game for an OS that is approaching the end of it's life cycle.

*Edit* Oh, and the 64 bit version of Windows XP was limited to SP2 only, so that means support has already ended for it.
 
PrinceofNothing said:
Well you'd lose that bet, because most of my knowledge comes from the fact that I've been a PC gamer for roughly 15 years now.

Now THAT is funny right there. I almost choked on my MET-Rx Protein Plus bar.
Stop embarrassing yourself.

Edit: actually, no. Keep going.
 
SystemShock7 said:
Now THAT is funny right there. I almost choked on my MET-Rx Protein Plus bar.
Stop embarrassing yourself.

Edit: actually, no. Keep going.

Stop embarrassing myself? Perhaps you should take your own advice. All you've contributed to this thread is to tell me to stop giving my opinion and to stop arguing with my supposed betters; despite the fact that CDPR did exactly what I said they would do out of sheer necessity. You haven't corrected me, or pointed out any mistakes that I've made in my assertions either.

Programming is complicated but it's not rocket science. Anyone with a decent IQ will easily be able to grasp the fundamental basics if they care enough to try.

So while it may be beyond your intellect, it's not beyond mine..
 
PrinceofNothing said:
Am I now?



DX11 has soft shadows by default, which can be done in DX9, but the performance cost is too great which is why developers never use it in DX9 mode.

Lighting? DX11 uses 64 bit HDR for more precise, and authentic looking lighting. DX9 is stuck with 32 bit precision, which doesn't look nearly as good.

Texture repetition? DX11 has a higher and more efficient texture compression ratio, which means they can use more detailed and varied textures without overwhelming the VRAM..

Better LOD? Never heard of tessellation I take it? Also, DX11 drastically reduces or eliminates object pop in when done correctly due to a more efficient rendering pipeline with less overhead.

My point is that you can have better graphics using dx9. There are dx9 games with WAY better shadows, lighting, AO, and great performance compared to TW2.

DX11 actually allows 128-bit HDR rendering but nobody use this. If you can't notice difference, why use it! Tessellation only make sense on objects that are really visible and not on distance objects.

Texture repetition is not big problem and can be fixed with different approach in design. The biggest problem here are games developed and optimized primarily for consoles. Not dx9.
 
Thor666Arise said:
Well I don't mind a DX 11 OPTION! but I really hope they don't make it exclusive for DX 11.
Yes it would look better no question but a lot of people with weaker graphic cards couldn't play it.
It would be a shame if those people couldn't enjoy the masterpiece that the Witcher 3 is hopefully going to be.
hey if you have a graphic card with no dx11 it so or so too week to play the game
a dx11 card costs not that much ... if you are a pc gamer 150 for a r9 270x should be ok
for physx 760 = 200 should be ok too
with any weeker then a gtx 500 it makes absolutly no sence too play a game with that graphics
it wont be fun to hang around with 20fps by low graphic settings and a very low res
so please make it dx11 only if there is anyone out there with no dx11 card why playing pc ?
 
BORCH3DOHLEN said:
hey if you have a graphic card with no dx11 it so or so too week to play the game
a dx11 card costs not that much ... if you are a pc gamer 150 for a r9 270x should be ok
for physx 760 = 200 should be ok too
with any weeker then a gtx 500 it makes absolutly no sence too play a game with that graphics
it wont be fun to hang around with 20fps by low graphic settings and a very low res
so please make it dx11 only if there is anyone out there with no dx11 card why playing pc ?

There are some of us with massive old graphics cards like GTX 285s that eat puny little 7770s and GTX 550s for breakfast. So your claim that nobody without a DX11 card should be doing gaming is offensive.

It would not be practical to demand that game developers continue to support old technology, though, so those of us with investments in these old beasts will have to be content with using them to play older games.
 
Ancient76 said:
My point is that you can have better graphics using dx9. There are dx9 games with WAY better shadows, lighting, AO, and great performance compared to TW2.

I agree. Witcher 2's strong points were it's texture quality and art direction. That's what made it look very good imo..

But the shadows, lighting, AO were only passing fair. Actually, the shadows were horrible to be quite honest.

DX11 actually allows 128-bit HDR rendering but nobody use this. If you can't notice difference, why use it!

I'm sure thats probably a matter of performance more than anything. 128 bit precision would result in a massive performance hit..

Tessellation only make sense on objects that are really visible and not on distance objects.

Not true dude. Ever played Battlefield 3 or 4? There's plenty of terrain tessellation in those games that make distant objects like mountain ranges or hills look more geometrically detailed from afar.

Witcher 3 will also use terrain tessellation as well to make the landscape more detailed, up close and from a distance. Water will be tessellated as well.

[media]http://www.youtube.com/watch?v=TBnl2vymRYs[/media]

Texture repetition is not big problem and can be fixed with different approach in design. The biggest problem here are games developed and optimized primarily for consoles. Not dx9.

I have to disagree. The first Witcher was infamous for having NPCs that looked the same. CDPR tried to remedy that in the Witcher 2, but it was still definitely present. I'm sure they'll have it in the Witcher 3 as well (although it should be greatly reduced), since it makes more sense to use texture variation for the different environments and locations than for peasants and other NPCs.

At any rate, I have a question for you.

Do you think the Witcher 3 could be made using DX9 and 32 bit, without sacrificing any of the attributes found in the DX11 and 64 bit path such as:

1) No loading zones

2) Complete contiguity with no chapters and artifically imposed boundaries

3) No, or greatly reduced LOD and object pop in

4) Proper multithreading (I don't know how well the Red Engine 3 is optimized for multicore processors, but I do know that DX11 makes it much easier to multithread your 3D engine)

5) More accurate and precise shadows, ambient occlusion (Witcher 3 will use HBAO+, the best and most accurate form of AO for games) and lighting.

Can all that be done in DX9 and on a 32 bit platform? I don't think so..
 
PrinceofNothing said:
I agree. Witcher 2's strong points were it's texture quality and art direction. That's what made it look very good imo..

But the shadows, lighting, AO were only passing fair. Actually, the shadows were horrible to be quite honest.



I'm sure thats probably a matter of performance more than anything. 128 bit precision would result in a massive performance hit..



Not true dude. Ever played Battlefield 3 or 4? There's plenty of terrain tessellation in those games that make distant objects like mountain ranges or hills look more geometrically detailed from afar.

Witcher 3 will also use terrain tessellation as well to make the landscape more detailed, up close and from a distance. Water will be tessellated as well.

[media]http://www.youtube.com/watch?v=TBnl2vymRYs[/media]



I have to disagree. The first Witcher was infamous for having NPCs that looked the same. CDPR tried to remedy that in the Witcher 2, but it was still definitely present. I'm sure they'll have it in the Witcher 3 as well (although it should be greatly reduced), since it makes more sense to use texture variation for the different environments and locations than for peasants and other NPCs.

At any rate, I have a question for you.

Do you think the Witcher 3 could be made using DX9 and 32 bit, without sacrificing any of the attributes found in the DX11 and 64 bit path such as:

1) No loading zones

2) Complete contiguity with no chapters and artifically imposed boundaries

3) No, or greatly reduced LOD and object pop in

4) Proper multithreading (I don't know how well the Red Engine 3 is optimized for multicore processors, but I do know that DX11 makes it much easier to multithread your 3D engine)

5) More accurate and precise shadows, ambient occlusion (Witcher 3 will use HBAO+, the best and most accurate form of AO for games) and lighting.

Can all that be done in DX9 and on a 32 bit platform? I don't think so..

All this can be done with OpenGL 4 as well, which works not only on Windows, but on MacOS X and Linux/SteamOS as well, which would increase possible userbase. So I completely disagree that CDPR must use DirectX 11. OpenGL 4 would be much better. Of course, they have more experience with DX, but according to Valve moving to OpenGL is not hard, so that would make a lot of sense for the future.
 
PrinceofNothing said:
I agree. Witcher 2's strong points were it's texture quality and art direction. That's what made it look very good imo..

But the shadows, lighting, AO were only passing fair. Actually, the shadows were horrible to be quite honest.



I'm sure thats probably a matter of performance more than anything. 128 bit precision would result in a massive performance hit..



Not true dude. Ever played Battlefield 3 or 4? There's plenty of terrain tessellation in those games that make distant objects like mountain ranges or hills look more geometrically detailed from afar.

Witcher 3 will also use terrain tessellation as well to make the landscape more detailed, up close and from a distance. Water will be tessellated as well.

[media]http://www.youtube.com/watch?v=TBnl2vymRYs[/media]



I have to disagree. The first Witcher was infamous for having NPCs that looked the same. CDPR tried to remedy that in the Witcher 2, but it was still definitely present. I'm sure they'll have it in the Witcher 3 as well (although it should be greatly reduced), since it makes more sense to use texture variation for the different environments and locations than for peasants and other NPCs.

At any rate, I have a question for you.

Do you think the Witcher 3 could be made using DX9 and 32 bit, without sacrificing any of the attributes found in the DX11 and 64 bit path such as:

1) No loading zones

2) Complete contiguity with no chapters and artifically imposed boundaries

3) No, or greatly reduced LOD and object pop in

4) Proper multithreading (I don't know how well the Red Engine 3 is optimized for multicore processors, but I do know that DX11 makes it much easier to multithread your 3D engine)

5) More accurate and precise shadows, ambient occlusion (Witcher 3 will use HBAO+, the best and most accurate form of AO for games) and lighting.

Can all that be done in DX9 and on a 32 bit platform? I don't think so..

Precision is a myth. It's too expensive. To make HDR more realistic people use methods like "tone mapping" and "bloom".

Tessellation in TW3 will not be used everywhere:

The biggest problem we have with tessellation is the way it creates holes in the mesh. This can be solved in two ways: one is to add a lot more data and a lot more calculations to the shaders, and the other is to change the assets in a way that avoids the problem. Both solutions are quite expensive, one in terms of performance, the other in terms of artist time. This is why we made the decision that using tessellation as a default for our characters wouldn’t give us as much as it would cost. That being said, tessellation can still be used with any mesh in the game because the system supports it – using it just becomes a decision that has to be made in some special cases.

We’ve done some experiments with adaptive tessellation; unfortunately, it would require quite a big pipeline change to really use it everywhere, plus we didn’t get that much benefit from it, so we decided that we will only use it in a few cases where the problems can be easily avoided and the difference is really visible.

Video that you posted only shows how tessellation can be applied to any mesh with their technology. They will use it only in some cases which is right decision.

Shadows in TW3 should not be extremely precise because this is third person game. So shadows aren't in front of your face like in the first person games.

For me the biggest benefit was introduced in dx10 with MSAA support for deferred rendering engines, and REDengine 3 is deferred.

You can also make a dx9 game using 64-bit platform.

On paper DX11 is great. But in reality things are quite different. The key to improve you graphics is to mix different methods like in case of HDR, or to mix tessellation with Parallax occlusion mapping for example.
 
I myself don't really care... as long as the story and gameplay is great then I will love it, if it looks anything like in the trailers then I surely won't be disappointed
 
jerf said:
All this can be done with OpenGL 4 as well, which works not only on Windows, but on MacOS X and Linux/SteamOS as well, which would increase possible userbase. So I completely disagree that CDPR must use DirectX 11. OpenGL 4 would be much better. Of course, they have more experience with DX, but according to Valve moving to OpenGL is not hard, so that would make a lot of sense for the future.

I have no issue with OpenGL, but realistically, the chances of CDPR doing a port is slim to none.

From what I've heard, OpenGL is harder to learn and to use than DirectX so if they have no direct experience with it, doing a port wouldn't yield the same kind of quality as the DirectX version.

You see this with Metro Last Light. The Linux version is totally inferior to the Windows version in practically every manner. Thats not an indication of the inferiority of OpenGL itself of course, but it's just that the devs just aren't as good with it.
 
Ancient76 said:
Precision is a myth. It's too expensive. To make HDR more realistic people use methods like "tone mapping" and "bloom".

It's because of the high precision that it's "too expensive."

Tessellation in TW3 will not be used everywhere

I never said it would be, but the game will obviously use it quite a bit, mostly for the environment. I think that excerpt you posted focused on character tessellation, and I agree with the developer that they don't need to use it for the character models as they are already very detailed.

Shadows in TW3 should not be extremely precise because this is third person game. So shadows aren't in front of your face like in the first person games.

I think you have it backwards. A third person perspective to me requires more accurate shadows than first person, as you're going to see them more often.

Besides, lighting and shadowing is extremely important for the game World at large when it comes to providing an immersive experience, and not just the characters.

You can also make a dx9 game using 64-bit platform.

True, but you couldn't get the kind of draw distances and reduction in object and LOD pop in that you get with DX11, because DX9 does not allow for parallel rendering unlike DX11.

On paper DX11 is great. But in reality things are quite different. The key to improve you graphics is to mix different methods like in case of HDR, or to mix tessellation with Parallax occlusion mapping for example.

To me DX11 is awesome. If you want to know how awesome, just look at the quality of the NATIVE DX11 games that have come out.

BF3, BF4, Assassin's Creed IV, Crysis 3, Civilization V etcetera not only look great, but run great as well. Witcher 3 will be another fine addition to the stable of that I have no doubt />
 
PrinceofNothing said:
It's because of the high precision that it's "too expensive."



I never said it would be, but the game will obviously use it quite a bit, mostly for the environment. I think that excerpt you posted focused on character tessellation, and I agree with the developer that they don't need to use it for the character models as they are already very detailed.



I think you have it backwards. A third person perspective to me requires more accurate shadows than first person, as you're going to see them more often.

Besides, lighting and shadowing is extremely important for the game World at large when it comes to providing an immersive experience, and not just the characters.



True, but you couldn't get the kind of draw distances and reduction in object and LOD pop in that you get with DX11, because DX9 does not allow for parallel rendering unlike DX11.



To me DX11 is awesome. If you want to know how awesome, just look at the quality of the NATIVE DX11 games that have come out.

BF3, BF4, Assassin's Creed IV, Crysis 3, Civilization V etcetera not only look great, but run great as well. Witcher 3 will be another fine addition to the stable of that I have no doubt />/>/>/>/>

You telling me that precision is to expensive after i told you and explained the same in previous post?? You really are something special.
There is no high precision in games. There are tricks, cheaper methods.

From the latest W3 interview:
"It didn't change that much in terms of poly-count or size of texture," lead character artist Pawel Mielniczuk told me - "it's like 30 per cent bigger than The Witcher 2, but it's nothing actually. Already there were in The Witcher 2 so many polygons on the character that you couldn't see the edges, so nobody cares about the polygons.

So poly-count is only 30% bigger. Amount of polygons on meshes will be only 30% bigger compared to W2.

You don't need tessellation to achieve realistic water:


Look how good and detailed mountains are without tessellation:


DX9 soft shadows:


Dynamic volumetric lighting and other features:

http://www.youtube.com/watch?v=IPVhQeG7WyY

You must experience this if you want to talk about it. I make mods including lighting, textures, models.
 
To the OP, I think the point most of the folks here are trying to make and you are consistently missing is that, just because something is 'new and improved' doesn't always mean it is better. Also there is a very old adage, that applies very much to technology, 'if it's not broke, don't fix it'. Something techs have a very bad history with, we always want to improve on things, that do the job outstandingly.

From what I've seen, and what the game industry has apparently proven time and again, is that the 'improvements' with DX10 were not great enough to alienate a large segment of their player base to step away from DX9, and right now, DX11 is still as of yet an unproven format. So it most assuredly falls under the 'if it's not broke, don't fix it' category. DX9 works beautifully - as many have already shown examples of here - can DX11 do more, most likely, however, is it currently worth the investment. My answer, most likely not. At this point in time, until it has proven itself further, I think it is a wiser decision to stay with what is proven, especially with the finale of a game series, than play with something that isn't.

That's my two Orens worth anyway, take it or leave it.
 
Ancient76 said:
You telling me that precision is to expensive after i told you and explained the same in previous post?? You really are something special.
There is no high precision in games. There are tricks, cheaper methods.

You need to follow the debate. You brought up 128 bit HDR and said that nobody used it because you can't tell the difference, after which I responded and said the reason they likely do not use it is because of the performance hit..

And that's when you started talking about it being "expensive."

So poly-count is only 30% bigger. Amount of polygons on meshes will be only 30% bigger compared to W2.

Yes, so they were already very detailed to begin with and would not benefit from tessellation.

You don't need tessellation to achieve realistic water

Using Crysis as an example of "realistic water" doesn't really help your argument. Crysis was the greatest back in the day, but since then, other games like Crysis 2, Crysis 3 and Assassin's Creed IV have much better looking water.

Crysis 2 and 3 uses tessellation for water, but Assassin's Creed IV doesn't. It just depends on whatever works best for the developer and their particular circumstance. Accordingly, CDPR will probably decide to tessellate the ocean in the game based on that tech trailer.

[media]http://www.youtube.com/watch?v=vRj8We3JpIk[/media]

Look how good and detailed mountains are without tessellation:

That shot was taken with streaming disabled, otherwise it would be a blurry mess. That's the whole point of tessellation. You can control the LOD of an object without using multiple textures.

DX9 soft shadows:

Nobody said DX9 couldn't do soft shadows. The reason why devs do not use that feature in DX9 is because the performance cost is too great, not because it cannot be done.

The Witcher 3 will be a vast open World game, so using DX9 with soft shadows would kill performance.

Dynamic volumetric lighting and other features:

DX10, which is a subset of DX11.

You must experience this if you want to talk about it. I make mods including lighting, textures, models.

Very good for you!
 
Kyriene said:
To the OP, I think the point most of the folks here are trying to make and you are consistently missing is that, just because something is 'new and improved' doesn't always mean it is better.

I think you're mistaken to imply that DX11 is "new." DX11 has been out for 4 years, and has been used in many games. DX9 on the other hand is positively ancient.

Also there is a very old adage, that applies very much to technology, 'if it's not broke, don't fix it'. Something techs have a very bad history with, we always want to improve on things, that do the job outstandingly.

Have you ever played a native DX11 game? Apparently not, if you think DX9 does an outstanding job..

From what I've seen, and what the game industry has apparently proven time and again, is that the 'improvements' with DX10 were not great enough to alienate a large segment of their player base to step away from DX9, and right now, DX11 is still as of yet an unproven format.

DX10 had massive improvements over DX9, and was responsible for breaking the old programming model. DX10 basically paved the way for DX11, which is a superset of DX10..

It's true that few developers used DX10/10.1, which is probably because it was such a radical departure for them and they needed time to learn it.

DX11 on the other hand, has been successful, with plenty of games utilizing it.

So it most assuredly falls under the 'if it's not broke, don't fix it' category. DX9 works beautifully - as many have already shown examples of here - can DX11 do more, most likely, however, is it currently worth the investment. My answer, most likely not. At this point in time, until it has proven itself further, I think it is a wiser decision to stay with what is proven, especially with the finale of a game series, than play with something that isn't.

You need to catch up with the times. You speak as though DX11 just came out last month or something, when it's been out for 4 years now.. There is NO QUESTION that DX11 can do more, as seen by the native DX11 games that feature not only much better looking and more sophisticated effects, but much better performance as well.

If DX11 isn't worth the investment, why is CDPR going full native DX11 with no DX9 fallback for the Witcher 3?
 
Top Bottom