The graphics

+
I kind of want to unify the debate here. The argument is NOT about direct illumination vs indirect illumination. There is no point debating what the shadows should look like, when there are none. Clearly the fire is a light source, multiple different materials in the scene are illuminated. Clearly the fire is NOT a source of shadow, there is not even the subtlest hint of any occlusion. It looks wrong.

No thats kinda my point, fire in daylight is not bright enough (compared to daylight) to cast visible shadows. Please check my pics of the real life burning cars above. The burning car is in daylight shadow and doesn't even manage to illuminate that, nor does the fire behind the truck make it cast a shadow in the other direction.

The pic with the motorbike headlights might need tweaking, but the fire - well i really start to think people forgot how weak normal fire is as a light source in reality (or how strong daylight)...
 
I'm gonna guess that shadows may still be glitchy / not optimized, so they turned them off. That way the demo would run well and look okay.
 

Guest 4310777

Guest
No thats kinda my point, fire in daylight is not bright enough (compared to daylight) to cast visible shadows. Please check my pics of the real life burning cars above. The burning car is in daylight shadow and doesn't even manage to illuminate that, nor does the fire behind the truck make it cast a shadow in the other direction.

The pic with the motorbike headlights might need tweaking, but the fire - well i really start to think people forgot how weak normal fire is as a light source in reality (or how strong daylight)...

I'm gonna have to meet you half way, I think the fire should be casting very subtle shadows. The scene is not very bright, perhaps under a bridge? And the fire is clearly bright enough to illuminate the environment with an orange glow. Now, keeping in mind that shadows are merely the ABSENCE of light, there should obviously be some obstructed areas where the orange glow doesn't exist. These obstructed (occluded) areas would be called shadows.

To put it another way, I don't want to see anything added to that scene, I want to see something subtracted.
 
I'm gonna have to meet you half way, I think the fire should be casting very subtle shadows. The scene is not very bright, perhaps under a bridge? And the fire is clearly bright enough to illuminate the environment with an orange glow. Now, keeping in mind that shadows are merely the ABSENCE of light, there should obviously be some obstructed areas where the orange glow doesn't exist. These obstructed (occluded) areas would be called shadows.
Sure, i would have no problem with that. However, isn't that already the case here? The highly reflective, polished car trunk to the left is only partially coloured by the yellow light, the other parts are illuminated by white daylight. Same for the puddles: They are showing strong reflections of the fire (like a mirror), just like the car. Or the white skylight reflections on the chairs arm rest vs. the yellow fire tint on the sitting surface. I'm actually amazed by how well even small variations are lit here, e.g. the light distortion on the car by the trunks curved surfave.

The white skylight reflections on highly reflective surfaces - asphalt is only highly reflective if wet - show the scene is illuminated by skylight quite strongly. Also its diffuse skylight. What i mean by this, the scattering is stronger than on a clear sunny day when its just atmospheric shattering by air molecules. As soon as the day is cloudy, or the air isn't 100% clear (smog, dust, fog etc), there will be less sunlight directly from the sun, but (if the clouds are not that thick = white) more indirect light those extra scattering sources - basically the whole cloud cover / smog serves as additional reflectors. Thats why you would get a strong shadow below a bridge on a sunny day (since most sunlight is direct) and a weak one without shadow contours (just general darkening) on a day with cloud cover.

Example of fire color tone but no visible shadow in daylight (also shows how only high reflection surfaces show color change unless extremely close, no fire tint otherwise):

RealFire.jpg


Many pics from fires are during the night where - absent other sources - the fire creates dramatic illumination and strong shadows. But on a cloudy day, or in direct sunlight, the extra illumination from fire won't even make a dent of a shadow (well physically yes, but not visible for humans). Maybe think of open fire while the other lamps are on in the living room, or a barbecue on a cloudy day...

To put it another way, I don't want to see anything added to that scene, I want to see something subtracted.
If the fire was pinpoint or small area like a torch and there was a obstucting object between fire and viewer, e.g. a big box instead of the chair, and it was close enough to the fire to be inside the area where the fire colors the (nonwet) asphalt, then the area behind that obstruction should not be tinted in the fire spectrum.

Sure, i would have no problem with that, thats correct. But adding a visible and contoured chair shadow from that gasoline fire - which is what some seem to expect, while strong daylight illumination (not direct sunlight, see above) is present - would not reflect reality. Might work as expected on a very clear sunny day, when the sun is directly above the bridge and everything below is in deep shadow.

I have no problem saying the graphics - especially in the deep dive video - still need adjustments. I'm just a little surprised that people took offense in the fire example specifically, cause that is more realistic than most other games or movie SFX fires i know.
 
Last edited:
Here's a factor I suspect people aren't considering with that scene/frame in which a few people have claimed 'no shadows'

It is under a bridge/over pass. The whole thing is in shadow, in which case there are just no deeper shadows - and that could be for a number of reasons.
 
I'm gonna guess that shadows may still be glitchy / not optimized, so they turned them off. That way the demo would run well and look okay.

This was my first though. Shadows are clearly missing, that cannot be intention. Like here - dark siulet on a well light surface, clear give away there is a strong light behind. But no shadow is cast towards the player as it should.
1567937040589.png


But strange thing is, shadows are there sometimes, like in 4:50.
Though somehow wierdly behaving (because the source is clearly somewhere in the middle of the corridor instead in the light or at the end).

And then there are those super short contact shadows
1567935560445.png


So my conclusion is, that some sources cast shadows and some does not. And some have just contact shadows. Optimalization as in any other game.
But what is strange is the distribution. Having headlight on a bike as not shadow casting is quite a wierd thing. And so having scenes with so many strong directional lights, but no shadow at all.
So I hope those are areas that are yet to be worked on by lighting artist, or that there will be an option for more shadow casting lights or something like that.

P.S. I would also want to remind people comparing this very specific situation to a game from 2008 that you can pick thing that were better in old games time to time. Like Half life had real time chrisp reflections on water surface but todays game usually does not (except Ray Tracing). That is tradeoff to todays more complex shaders and geometry and so such situations needs to be compared as a whole.

P.S.S This pixel hunting in shown material before actual realease is a terribel thing anyway. :-D
 
Graphics, as intensive as they are, will usually be among the first things to get scaled back to be sure the rest of the game functions well.

That's what optimization is.

Correction: that’s what “optimization” is these days. Back in 2004 it meant fixing errors in code and optimizing it not cutting visual features left and right.

Compare Half Life 2 and Dark Messiah of Might & Magic performance: the former is running great and the latter runs like crap. Both are using the same engine yet HL2 is optimized with no visual concessions.
Post automatically merged:

Juist noticed, the deep dive video has this in the description.

View attachment 11014331
Well there you go, so much for the "downgrade" debacle, its as if people hate doing research before jumping to the gun.

Since when did YouTube compression algorithm get so sophisticated as to remove character shadows and make overall lighting flat? P.S. The Witcher 3 marketing campaign used the same ploy i.e. YouTube removed particle effects, volumetric lighting etc.
 
Last edited:
Since when did YouTube compression algorithm get so sophisticated as to remove character shadows and make overall lighting flat? P.S. The Witcher 3 marketing campaign used the same ploy i.e. YouTube removed particle effects, volumetric lighting etc.
No, its not responsible for those. But since we have the comparison to the 3.8GB video, and there is a definite difference in texture quality between 1080p on the video download vs. 1080p on twitch/youtube. So the texture quality difference (or rather extra blur) between those two is indeed youtube compression.

However, for some of those light sources i'm not sure whether shadows are "missing" at all: We are so used to the pinpoint lightsource shadows (clear cut shadow) used in most games, that the actual effect on diffuse light areas on shadows in reality seems to be almost forgotten. I don't have a huge diffuse area light to show it perfectly, but even the difference of the same lamp with a globe (=light from area) and without a globe - no other light source - has a very significant effect on object shadows (or their absence) in realty.
DiffuseShadowsExample.PNG


Not saying this is the reason for all "missing" shadows, but e.g. Wenceslaus posted an image of a gang member in front of a diffuse lightning area (soda machine) with only contact shadows. Which is physically correct, there would be no defined shadow behind the gang member....
 
Last edited:
At this point, I hope CDPR release potato graphics gameplay reveal from now on. People will be grateful for the upgrade they got when the game comes out.
 
Back in 2004 it meant fixing errors in code and optimizing it not cutting visual features left and right.
Knowing that there's not many games in the history that showed you 50 mins of 2 years early WIP version of the game let that be 2018 or 2004, that's some next level insider knowledge from you man.
 
I know this is a CDPR forum but goddamn there are so many apologists in here it's crazy.

[...]

Just rewatched the 15 min supercut and remembered something tacky and cheap, namely flashing soda cans and cash stacks. Here’s to hoping you can turn the flashing off.

Also I wonder if the UI change is due to previous UI lead, Alvin Liu, leaving the studio.

[Edited -- SigilFey]
 
Last edited by a moderator:
Just rewatched the 15 min supercut and remembered something tacky and cheap, namely flashing soda cans and cash stacks. Here’s to hoping you can turn the flashing off.

Also I wonder if the UI change is due to previous UI lead, Alvin Liu, leaving the studio.
This will probably be the first CDPR game I play for more than 5 hours. The Witcher wasn't my genre but I gave it a shot. Story couldn't pull me through the game given the genre and combat. So I'm coming into their games pretty fresh and without the "trust" that so many seem blindly placed.

Call me cautiously optimistic. I'm sure the game will be good given the genre and style, but there are already things I think holding it back from being great
 
Last edited by a moderator:
This will probably be the first CDPR game I play for more than 5 hours. The Witcher wasn't my genre but I gave it a shot. Story couldn't pull me through the game given the genre and combat. So I'm coming into their games pretty fresh and without the "trust" that so many seem blindly placed.

Call me cautiously optimistic. I'm sure the game will be good given the genre and style, but there are already things I think holding it back from being great

I still think The Witcher (the first one) was the best game CDPR ever made but it was a different studio back then. It was unique in its atmosphere and adherence to the source material is unparalleled. Two sequels fall more in a typical Western action adventure territory.

I also think past merits are not a good gauge of quality of a future product because the staff in gaming studios changes frequently so that many people that made your favorite game might no longer be among its ranks. Look at BioWare.
 
Top Bottom