Crytek and unreal 3 demo...which one has better graphic
- Thread starter beginner16
- Start date
beginner16
Newcomer
XxStratoMasterXx said:
Can you give me an example of where doom 3 engine is superior to unreal 3 engine?
Ratchet said:
Actually, that's wrong. That "example" link is NOT generated by the Doom3 engine. It is a scene someone would like to see them generate using the D3 engine. The sample shots that are actually generated by the D3 engine don't seem that impressive so far.
Taken within context, one can see this is merely links to scenes they'd like to see generated:
MasterBaiter
Regular
Did Valve or Crytek use motion capture for either of their games out curiosity? 
The motion in the Crytek demo does seem stunted.
The motion in the Crytek demo does seem stunted.Here though this is from Doom 3 and looks nice
I wish I could find the thread it was in, he had better pictures as well, much better, but I could not for the life of me find it again...
By the way doing the Euro looking thing (they are tryig trafalgar square) would be easy if they did what BNA! suggested, make a model out of the building put a normal map on it and it will look like the flat HL2 buildings that you couldn't enter.
I wish I could find the thread it was in, he had better pictures as well, much better, but I could not for the life of me find it again...
By the way doing the Euro looking thing (they are tryig trafalgar square) would be easy if they did what BNA! suggested, make a model out of the building put a normal map on it and it will look like the flat HL2 buildings that you couldn't enter.
XxStratoMasterXx
Regular
beginner16 said:
Unified Lighting model.
Leto said:
Which part in particular are you referring to?
beginner16
Newcomer
XxStratoMasterXx said:
I only have faint idea what you mean by that,but I find it kinda weird that a game engine made for cards in one year time from now,completely rewriten for DX 9 would be less advanced in any way than engine written by one person(carmack) two years earlier.I hope you 're wrong
The concept is a step forward, but the implementation leaves much to be desired.XxStratoMasterXx said:
The whole point of a unified lighting model is uniformity in shadowing (lighting will depend on the material, so you don't want it uniform). However, I do not believe this is necessary at this point. It is enough to make shadowing techniques similar in apparent results, and completely automatic and transparent to content creators.
XxStratoMasterXx
Regular
Ostsol said:
What's wrong with the implimentation? Everything in the game is lit the exact same way, and everything is dynamically shadowed, the exact same way.
Oh, and Unified lighting and shadowing doesn't mean everything looks like plastic...that's just due to the fact that JC and the crew at id chose not to utilize fragment programs for the core engine, and use varied tighter and broader specular highlights, due to wanting to keep the experience the same on all cards.
XxStratoMasterXx
Regular
Reverend said:
In what respect? Technologically, i'd agree with you. However as much as I worship JC and id software, I think that Unreal Engine 3 demo that was shown at E3 and GDC and the GF6 launch party...well smokes D3 in terms of art and content
XxStratoMasterXx
Regular
beginner16 said:XxStratoMasterXx said:
I only have faint idea what you mean by that,but I find it kinda weird that a game engine made for cards in one year time from now,completely rewriten for DX 9 would be less advanced in any way than engine written by one person(carmack) two years earlier.I hope you 're wrong
The ARB2 and Cg paths in Doom 3 are effectively DX9 only renderers.
There is no need to have all shadows dynamic. I understand how it can be desirable in order to have all shadows cast properly onto dynamic objects, but stencil shadows make this very inefficient. Scenes will become vertex limited (due either to hardware vertex processing limitations or to CPU speed) much faster than with other methods that can achieve the same effect. There is also the issue that the shadows are all hard-edged, which is okay for short shadows in relatively confined areas, but not for large expanses.XxStratoMasterXx said:What's wrong with the implimentation? Everything in the game is lit the exact same way, and everything is dynamically shadowed, the exact same way.
Oh, and Unified lighting and shadowing doesn't mean everything looks like plastic...that's just due to the fact that JC and the crew at id chose not to utilize fragment programs for the core engine, and use varied tighter and broader specular highlights, due to wanting to keep the experience the same on all cards.
The materials -really- could have been done better. Making things look the same across all hardware is a poor excuse. I paid for a Radeon 9700 Pro not just for speed, but for the technology it supports. I expect games that have been in development for so long to use more than a couple shaders for every material. In Doom 3 it's not often noticable, mainly because everything is so dark; but in brighter lit areas near the end, all the stone began looking like plastic, to me.
Luminescent
Veteran
Well put, Ostsol.
XxStratoMasterXx
Regular
Ostsol said:There is no need to have all shadows dynamic. I understand how it can be desirable in order to have all shadows cast properly onto dynamic objects, but stencil shadows make this very inefficient. Scenes will become vertex limited (due either to hardware vertex processing limitations or to CPU speed) much faster than with other methods that can achieve the same effect. There is also the issue that the shadows are all hard-edged, which is okay for short shadows in relatively confined areas, but not for large expanses.XxStratoMasterXx said:What's wrong with the implimentation? Everything in the game is lit the exact same way, and everything is dynamically shadowed, the exact same way.
Oh, and Unified lighting and shadowing doesn't mean everything looks like plastic...that's just due to the fact that JC and the crew at id chose not to utilize fragment programs for the core engine, and use varied tighter and broader specular highlights, due to wanting to keep the experience the same on all cards.
The materials -really- could have been done better. Making things look the same across all hardware is a poor excuse. I paid for a Radeon 9700 Pro not just for speed, but for the technology it supports. I expect games that have been in development for so long to use more than a couple shaders for every material. In Doom 3 it's not often noticable, mainly because everything is so dark; but in brighter lit areas near the end, all the stone began looking like plastic, to me.
Well the materials isn't so much an engine problem as an artistic problem.
XxStratoMasterXx
Regular
Well it was this stencil shadow unified implimentation that is now paving the way for JC to do a SOFT shadow everywhere implimentation.
I think that JC was 100% correct to go this route for a paradigm shift...however if a developer doesnt need a premium engine that is very advanced, then a dev can licence Unreal Engine 2, Source or CryEngine...
I think that JC was 100% correct to go this route for a paradigm shift...however if a developer doesnt need a premium engine that is very advanced, then a dev can licence Unreal Engine 2, Source or CryEngine...
I don't think stencil shadows really have anything to do with the move to the next shadowing technique he might use. IIRC, the next technique will be completely unrelated.
Just some thoughts:
I'm pretty sure that it'll be some derivative of depth-buffer shadowmaps. The one problem people attribute to them is that it doesn't work all that well for pointlights. The reason for this argument is that one needs to render six shadowmaps per-frame per-light, one for each side of a cubemap, in order for the light to project shadows in every direction. This certainly is necessary for moving light sources, but is totally unecessary in the majority of cases where the light is static.
For static light sources, the shadowmaps can be pre-generated and stored in memory. Each frame the engine will find out which cube faces are necessary to cast dynamic objects' shadows and will render those dynamic objects (but not the static ones) to new shadowmaps. The result will be combined with the static shadowmaps and the results will be projected.
The beauty of this is that unlike with shadow volumes, each pass can render several lights.
Just some thoughts:
I'm pretty sure that it'll be some derivative of depth-buffer shadowmaps. The one problem people attribute to them is that it doesn't work all that well for pointlights. The reason for this argument is that one needs to render six shadowmaps per-frame per-light, one for each side of a cubemap, in order for the light to project shadows in every direction. This certainly is necessary for moving light sources, but is totally unecessary in the majority of cases where the light is static.
For static light sources, the shadowmaps can be pre-generated and stored in memory. Each frame the engine will find out which cube faces are necessary to cast dynamic objects' shadows and will render those dynamic objects (but not the static ones) to new shadowmaps. The result will be combined with the static shadowmaps and the results will be projected.
The beauty of this is that unlike with shadow volumes, each pass can render several lights.
Similar threads
