Witcher 3 downgrade arguments in here and nowhere else

Status
Not open for further replies.

silva1991

Member
Now look at this infamous comparison image:

As it is (which is slightly unfair because one is high bit-rate and the other isnt):
cxi41b3g7ul5.jpg


vs.

Colour-shift and sharpening:
cxi41b32_3qdubn.jpg

The older one still looks more natural and better.

you did make the different look less significant so I'll give you that.
 

tuxfool

Banned
The older one still looks more natural and better.

you did make the different look less significant so I'll give you that.

He is still sharpening a low resolution youtube screenshot, not a 4k direct feed capture.

It should also be noted that those captures don't seem to be from the same location.
 

SaberEdge

Member
Not to pick your wording apart, but I seriously think there is a level of sharpening going on which people are mistakening for higher detail or higher quality.

For example look at this screen from the Sword of Destiny trailer:

Look at the darkened edges around each blade of grass (although it is just part of one sprite.

Now look at your screen you just posted:


Now look at a sharpened image of that screen:


What do you think or notice there in the second image in relation to the way vegetation looks? I personally think the vegetation with a sharpening filter is eerily reminiscent to this "higher quality" vegetation from previous TW3 media.

I think you're right. That does seem to improve the look of the ground foliage in screenshots. But I also suspect that such sharpening wouldn't be necessary or even desirable when playing the game on our own displays.
 
When I mean Foliage, I want to see something like this
*Screen from DriveClub.

Of course, TW 3 being an open world game doesn't have to render the best foliage, but it could have been better. I mean Skyrim Mods look pretty damn good compared to what we have in TW 3, and it is also an open world game.

I am not really sure how to explain it in technical terms, but if you go back a few pages, someone did share a representation of how the foliage in TW 3 looks and feels like. I don't know about you, but I do find it a step back from the original reveal.
I will always agree that rendering could be better done (even though vegetation rendering is done better elsewhere than in DC, but it is a good example of high resolution textures). That is completely valid and encouraged. Not every game is Crysis 3 or Ryse in terms of vegetation, and even those games could do it better.

I just think this case as "downgrades" are is different. The term is not warranted here regarding the grass rendering (it was always sprites that were kinda poor looking, even in the old trailers). I think it is just a case of the art direction for it specifcally getting worse in dev. That is all.
 
Now look at this infamous comparison image:

As it is (which is slightly unfair because one is high bit-rate and the other isnt):
cxi41b3g7ul5.jpg


vs.

Colour-shift and sharpening:
cxi41b32_3qdubn.jpg
The colour shift certainly does make it look similar, but you still can't ignore that the vegetation is overall lower resolution and overall variety.
 

VE3TRO

Formerly Gizmowned
Even if the shots from the PS4 were taken from the same time of day the game features a dynamic weather system.
 
The colour shift certainly does make it look similar, but you still can't ignore that the vegetation is over lower resolution and overall variety.

The newer image is lower resolution which makes this comparison so damn skewed. In fact, the best would be to find an area in retail with similar vegetation density and variety.

Of course that 1/4th area of a screen has less variety. It is also a a cheerry picked example to prove an argumentative point (and a baseless one at that).
The only thing this screenshot shows me is that the're just randomly clicking on empty spots of terrain to place a few grass sprites. But it looks very unnatural...

The game is open world and used a procedurally placed grass system. I can hardly fault them for having poorly placed grass in certain scenes. But yeah, it doesn't look that great at times. I am fine with saying that. It being a technical downgrade? No. I do not think that is a good argument.
 

viveks86

Member
But is the older one technically more demanding aka. requiring a downgrade to make it run at good framerates (on console)?

That is my entire point. People use the word downgrade to encapsulate TOD changes and colour palette differences... when it should only be a technical category. I am completely fine with you liking one more than the other.

Also, btw, these use the exact same technology. Just gemoetric planes with a sprite. One may be a better made sprite than the other artistically.. and placed more "painterly." But that does not make it more technically demanding and hence requiring a downgrade.

The density of said sprites could be demanding. The original certainly has much higher density. May be it was too demanding for the amount of collision detection they were going for, in addition to sheer number of draw calls. Besides, we don't really know if the technology behind them has changed in order to accommodate for some performance bottleneck. May be the older ones were procedurally generated (thereby requiring more processing) whilst the newer ones were manually 'brushed' over. That could explain why the newer sprites seem to repeat itself (in the same scene) more than the older one. It's all up in the air at this point and we really need more 1:1 comparisons. The wheat field comparison sheds some light on it, but the vegetation there has been given such a complete overhaul that the comparison seems pointless. I'm sure we'll find other areas that would be more apples-apples.


What about this one?
QcOH0nT.png

That particular one is invalid. Let's stop comparing with images that were taken down. It's a waste of time. For all we know, that was an early WIP shot and someone screwed up.
 

RedSwirl

Junior Member
After all that time people spent trying to get CDPR to get rid of the terrible sharpening filter and over-done colour grading, it turns out a great deal of GAF loves this stuff and mistakes it for actual tangible technical changes in a game (worse texture resolution, different shading model).

A number of GAffers in this thread misuse the word lighting. People using it on average just mean colour palette... not actual technical details related to lighting and shading.

Apparently these images are cartoony dragon age wasteland:


And these are hyper realistic nextgen:


I am fine with people arguing this topic, but please argue points of actual technical merit and understanding.

This is the post everybody needs to look at.

Nobody's denying 2015 Witcher 3 looks very different from 2013 Witcher 3, but a lot of people seem to be mistaking a change in art direction for a linear downgrade in tech.

From what I can tell, CDPR de-sharpened the game and made a huge change to the color palette. A lot people seem to think the sharpened look is objectively "better" or "more atmospheric" and think the new look is "flat," but people hated the sharpening back in Withcer 2. Same goes for the colors. A lot of the old screenshots and videos had some very deep colors that popped out. Now CDPR seems to have equalized things for a more natural image, and people seem to think that looks "flat" as well.

In my mind I connect this to a lot of the ENB Skyrim screenshots I see. A lot of people really seem to like throwing intense colors and other effects like bloom and DOF all over the place when they mod games. Maybe it's a "more is better" mentality when it comes to pushing graphics. Maybe people are just really used to a sort of "video game" look.

I just think this whole argument is really a matter of preference. Some prefer the 2013 look, but are arguing it looks objectively better. This is nothing like Dark Souls 2 where From Software straight-up cut a whole dynamic lighting system from the game. The Witcher 3's lighting was just changed, not removed.

And yes, assets were definitely changed and replaced, but whether they look "better" or "worse" is almost entirely subjective in what I've seen. The only area where I think you can argue an actual quality change occurred is in the foliage LOD, and even that was a partly artistic change if you ask me.

Nope. The one also uses flat planes. There is more vegetation, and mixed height so it isn't as visible, but it is the same.

I agree with this post. The foliage assets are definitely different now. The old foliage has more height variation and seems to overall use a higher variety of assets. It still looks entirely flat and not polygonal to me. Even in the very earliest trailers I thought all the foliage was flat, but it just happened to show really well in those cherry-picked shots.
 

Pop

Member
Thats some crazy magic. :D

Can you pleas do this with this pic if it is not to much work. It has the same flowers like the one in your comparison. I would love to see how it would look.


10856558_1596112283965645_6814697985301896742_o.jpg

Not all grass is created equal.

Can't believe I had to address that.
 

tuxfool

Banned
It is but I was following the labels on the site for reference.

For example, this:


Has the sharpening filter pretty clearly.

But I'm not sure the same is applicable here:

Ah yes, you're right. Those SoD screens seem to be native resolution to the final size. They're lacking in AA.

The first image is supersampled and then sharpened. The Supersampling produces much nicer results on alpha. See this shot from DA:I. Granted, Frostbite has crappy MSAA but you get the idea.

http://screenshotcomparison.com/comparison/104275
 
The density of said sprites could be demanding.
I think we have to wait for the game to come out to compare those things in the exact areas where the trailer was made (like you said yourself). This thread stinks of unequal comparisons, and I want to get rid of that. If the end game proves to have less density and variety in the end (whilst supersampling and perhaps sharpening the comparison images), then yes, I would happily agree there was a technical downgrade.
The original certainly has much higher density. May be it was too demanding for the amount of collision detection they were going for, in addition to sheer number of draw calls. Besides, we don't really know if the technology behind them has changed in order to accommodate for some performance bottleneck. May be the older ones were procedurally generated (thereby requiring more processing) whilst the newer ones were manually 'brushed' over. That could explain why the newer sprites seem to repeat itself (in the same scene) more than the older one. It's all up in the air at this point and we really need more 1:1 comparisons. The wheat field comparison sheds some light on it, but the vegetation there has been given such a complete overhaul that the comparison seems pointless. I'm sure we'll find other areas that would be more apples-apples.
I think you have it backwards. The older ones would have been hand placed (hence their artistic look), newer ones, after the world has been fleshed out in size, could only realistically been placed in a procedural manner.
 

tuxfool

Banned
I think we have to wait for the game to come out to compare those things in the exact areas where the trailer was made (like you said yourself). This thread stinks of unequal comparisons, and I want to get rid of that.

I've been railing against this fact the whole thread. Can't stop a moving train.
 

Dr Dogg

Member
What on earth am I reading. To people dismissing this comparison as pointless or the footage wasn't a match what are you even looking at?

While the first image in from the January 2015 footage (if I recall correctly) the second is apparently for the retail PS4 version which if you look at the whopping great areas I've circled you can clearly see the geometry has changed from a less complex wall to a one with a more complex model resulting in individual bricks being apparent. Doesn't make a blind bit of difference what time of day, resolution or quality and compression of screen capture that there is clear as day.

Right now this thread has a lot of people arguing over aesthetics rather than anything of technical merit which I think some would be better severed realising the difference.
 

OraleeWey

Member
It is pretty clear that it is downgraded. But why don't you guys expect it? We should all know by now that every game gets a "downgrade" after the reveal and leading up to release. .
 
I think we have to wait for the game to come out to compare those things in the exact areas where the trailer was made (like you said yourself). This thread stinks of unequal comparisons, and I want to get rid of that. If the end game proves to have less density and variety in the end (whilst supersampling and perhaps sharpening the comparison images), then yes, I would happily agree there was a technical downgrade.

A-fucking-men
 

stabiliser

Neo Member
Not all grass is created equal.

Can't believe I had to address that.

I don't understand? Are you some sort of grass rasicst or what? Joke aside, i just want to see the yelleow flower with the sharpening filter, i won't mind if you say no.
 
But is the older one technically more demanding aka. requiring a downgrade to make it run at good framerates (on console)?

That is my entire point. People use the word downgrade to encapsulate TOD changes and colour palette differences... when it should only be a technical category. I am completely fine with you liking one more than the other.

Also, btw, these use the exact same technology. Just gemoetric planes with a sprite. One may be a better made sprite than the other artistically.. and placed more "painterly." But that does not make it more technically demanding and hence requiring a downgrade.

And I hate to abuse page order... but I am requoting myself:

Hi-res and dense sprites take more power. Just look at the X1 version of GTA5.
 

Septic360

Banned
What on earth am I reading. To people dismissing this comparison as pointless or the footage wasn't a match what are you even looking at?


While the first image in from the January 2015 footage (if I recall correctly) the second is apparently for the retail PS4 version which if you look at the whopping great areas I've circled you can clearly see the geometry has changed from a less complex wall to a one with a more complex model resulting in individual bricks being apparent. Doesn't make a blind bit of difference what time of day, resolution or quality and compression of screen capture that there is clear as day.

Right now this thread has a lot of people arguing over aesthetics rather than anything of technical merit which I think some would be better severed realising the difference.

What about le face! LE FACE!
 
I don't understand? Are you some sort of grass rasicst or what? Joke aside, i just want to see the yelleow flower with the sharpening filter, i won't mind if you say no.
I didn't mask out geralt this time (takes too much time).
But here you are:
Normal
Sharpen and colour-shift
Hi-res and dense sprites take more power. Just look at the X1 version of GTA5.

I am going to wait for retail to see whether or not their resolution is so different, or whether their density and variety is so "downgraded". In the mean time, I think I have proven quite well that a colour change and a sharpening filter is the majority of what people are seeing in these rather non 1:1 comparisons.
 

Trace

Banned
Can't believe people are trying to compare textures from such poor quality screenshots.

It amazes me people will take 360p screenshots from different angles, time of day, even different versions of the game, and then try and compare with existing screenshots. Until I see some actual decent screenshot comparisons I fail to see the "massive downgrade".
 

tuxfool

Banned
While the first image in from the January 2015 footage (if I recall correctly) the second is apparently for the retail PS4 version which if you look at the whopping great areas I've circled you can clearly see the geometry has changed from a less complex wall to a one with a more complex model resulting in individual bricks being apparent. Doesn't make a blind bit of difference what time of day, resolution or quality and compression of screen capture that there is clear as day.

The PC capture is from a video released by CDp. People complained about the bricks so they increased the geometry. The PS4 build is the latest build, whereas the PC build is back from Jan.
 

SaberEdge

Member
Now look at this infamous comparison image:

As it is (which is slightly unfair because one is high bit-rate and the other isnt):
cxi41b3g7ul5.jpg


vs.

Colour-shift and sharpening:
cxi41b32_3qdubn.jpg

Nice job on that. Aside from the fact it's somewhat denser in the older shot (considering they aren't even comparing the same place) I'd say you got the base short grass to look fairly similar between the two. Add in some of the yellow flowers and taller wheat-looking grass, which we've definitely seen in the newer footage, and I don't think the overall look would be that different.

3Ufa0sL.jpg

Yellow flowers example.

YH4No7e.jpg

2hokILD.jpg

hBbBRvA.jpg

Wheat-type examples.
 

DieH@rd

Banned
A downgrade occurs because things are computationally expensive. Everything I see about this grass looks the same to me in terms of why it would be a certain level of GPU or CPU time.

Older shot has more types of grass, and each bush is smaller and made from higher res textures.There is more of them on screen, and in windy conditions, they all create illusion that nice looking grass is moving convincingly.

Newer shot has less brushes. Each of those bushes has more blades of low-res grass [larger sprite with low-res blades of grass], and in windy conditions less sprites are moving, creating poorer representation of real grass.

Newer shot is less computationally heavy because less stuff is processed and shown on screen, but it also looks worse. All changes we noticed so far [foliage, LOD, lightning and particles] were made for that reason, to be less computationally heavy.
 
Status
Not open for further replies.
Top Bottom