some weeks ago I got the idea of making an owm udk based project. In contrast to the majority of such projects, which are usually games or showcases, I decided to join 2 fields of my interest and making a benchmark.
Primary objective is to create a level that provides very good visuals - this includes the use of up to date features like tesselation (dx11) or parallax occlusion mapping (dx9). The setting will be somehow medieval like, inspired by certain rpg's.
It took me some time to learn more basic stuff about 3d modeling as well as getting better experience and routine. Of course, the process of learning never comes to an end but I feel beeing ready to get it started properly.
For now the project is in it's very early days and I'm still looking for some fellows to form a little team of 3-5 members.
Some days ago I began to model a first part of the level mesh (I don't feel comfortable with ukd's terrain/landscape system):
Also here's an older screenshot of some tesselation displacement mapping:
-- Edited by mAlkAv!An on Sunday 10th of June 2012 09:29:54 PM
When I first launched UDK I was blown away by its graphics and the tessalation examples. I was like "That's like bumpmapping's next level! This looks so 1337!" (I mean the lava texture on the sphere static mesh which was provided as an example).
I'm curious how your little benchmark project will look like and how it will run on my PC.
-- Edited by Sly. on Wednesday 11th of January 2012 10:37:35 PM
Sounds interesting, but I'm not sure what you mean. Is it that you wish to create a level so you can perform benchtests? If so will the levels test a specific bit of technology to assess how much effect it has or is it more generic system performance?
Even though I'm not the maker, let me try to explain it: The latest UDK version has (of course) the latest UE3 version (if Epic says the truth, possible that they have a better one that they keep secret and will use it in one of their upcoming games and won't release it before or something of that kind *rolleyes*). With it come plenty of new effects, the required specs changed and graphics overall are better.
Now, if you are curious whether you can run one of the latest version UE3 games, just run the benchmark (which will have a matinee flythrough and show cirtical places) and judging by the average of frames which you can show with stat fps you will know how smooth such a game (which is pushed to the limit that is - keep in mind that many have lower requirements because they adjust them to consoles too - you see, consoles can have a good side ) will run. Since a matinee guarantees the same "measurement" for each PC it's launched on, it can be a reliable method to see how a latest version UE3 game will run. In fact, I've also considered doing something like that. I have made a nice Japanese garden in UDK as a testmap (dislike the selection and movement of actors... I keep moving them accidentally :/ but at least the effects are "Wow!") and thought about doing that but Malk's version of a benchmark seems by far more interesting to me.
Interesting idea. Art and tech testing in one package. I remember the Crysis benchmark which was a cool thing to use to test how your software tweaks and hardware changes made a difference.
Thanks for the replies. Sly. has explained it pretty well, just think of other benchmark progs like 3dmark or unigine heaven for example. It's not intended to make this a playable level but there will be a defined autmatic flytrough. Afterwards fps stats (and perhaps also some kind of a score ) will be displayed to compare your system.
If someone of you is interested to help please let me know. For now it's all about modeling/texturing but at an advanced stage there will also occure questions about level design (apart from the base level mesh).
-- Edited by mAlkAv!An on Friday 13th of January 2012 09:16:00 AM
If you get to the stage of level design you could ask me if you like, there is a slight chance I would have some time to help you out, can't help much with modelling nor texturing atm and I still have a map to finish.
Well as the purpose is just to create a benchmark specific for gaming the design questions are purely form / aesthetics only. But it sounds interesting if it can quantitatively answer questions about how much a specific component (level tech) can impact performance to better understand the level optimisation.
If happy to let me use the other grasses shoot them my way.(especially that one with the taller grass with small bulbs at the top - on the left side of the pic)
Otherwise if you wanna keep them exclusively for ya benchmark map first then no probs M8!
Keep up the good work!
-- Edited by Lord_PorkSword on Tuesday 28th of February 2012 11:29:38 AM
Started to do some buildings... here's a simple wooden house. A second version with different door/window placement will follow. I also began to make a medieval windmill (bock mill style).
Textures are not yet final.
perspectives:
Lods:
in action ^^
-- Edited by mAlkAv!An on Tuesday 17th of April 2012 01:25:02 PM
Mick, this is intended to be a simple house, thats why there's will be only a campfire in the open air. The more advanced houses will have a fireplace inside and a chimney.
-- Edited by mAlkAv!An on Tuesday 17th of April 2012 06:15:16 PM
lol'd at the third picture. Is that the Demo Guy leaning out of the window? So that's where the dude was hiding since half a year already, long time no see Demo Guy (bug)! ROFL
Nice meshes you did there, imo it has a celtic feel to it.
Regarding the subject matter these are mainly for some woodcutter huts. I'll might move on with a hunter, stonemason, the windmill or make some town buildings first.
-- Edited by mAlkAv!An on Monday 7th of May 2012 08:26:11 PM
I was wondering if upon close up one can see, crystalline flakes or is it just particulate?
The other thing I wondered was as the snow falls will you be implementing any transitions to opacity for the ground and tops of buildings / props as they are covered in snow? Tricky to pull off but will really sell the effect.
Keep up the good work, looking forward to the progress on this one.
The snow itself is just preliminary. I will use a particle system later on, atm it is a mesh made out of multiple cylinders with a panning texture. The texture is just some random blobs and dots.
As for the transition I'm not completely sure what you mean. There is a transition between bare areas of an object and those covered by snow, taking the normal map into account - but it's not working across hard edges and large angles.
Ok, got your point. It's possible to vertex paint the areas, though this would require lot's of work for a whole enironment and you'll have to place each object in a predefined way. At the moment it's all in the material, no vertex colors or texture masks. The shader blends in snow based on the world up vector, so it does only affect surfaces being horizontally facing upwards +-90°. If i rotate e.g. the stone by 180° it's former bottom will get the snow and the former top has the stone texture. (like shown here eg) If you watch the video closely at the end you'll notice that the whole ground is growing already, that's done by a simple vertex shader.
-- Edited by mAlkAv!An on Thursday 30th of August 2012 04:04:39 PM
Yeah I'm not sure entirely. I was thinking it could be implemented using weight painting on the mesh. Alternatively one would have to create a separate UV layout to mask the texture. If that didn't work I guess using swapping out the exposed faces of the mesh with a custom mesh of the exposed faces and procedurally layer them onto the scene transitioning the growing piles of snow. One could also imagine using a draw scale 3d to enlarge the new mesh with the snow texture on it to simulate the growing piles of snow.
I'm running HD 6850, but I've never scored high on any benchmark in the past so I will remain sceptical of my pc's performance, for the time being. ;) What sort of things would you like to do with dynamic lighting? I'm still reading up on what UT3 and UDK can do in that respect, most fascinating.
-- Edited by Winter Dragon on Wednesday 5th of September 2012 04:27:05 AM
What's your gfx card? Since I will (also) use fully dynamic lighting -which does not run well with UE3- it might be a hard task for older/lowend cards indeed. But I'm trying to keep it at a playable frame rate (aiming for 30FPS+ at 1920x1080px, DX9) with mid classes cards like ATI HD5850/6850/6870 or Nvidia GTX460/560/Ti.
I'm going for a day/night-cycle, that's why everything needs to be dynamically lit. Most benchmarks are not based on a real game engine and there are lots of things to tweak to get a good score, plus the CPU speed is often very important, too. I guess if you run 3dmark01/05/06 or aquamark3 with your HD6850, I can show you a similar (if not better) result that I've made with a gfx card that is several years older and performs much worse in real games just because of those reasons.
This cave scene is for testing multiple dynamic lights at once. I'm working on the outdoor area now which will come with a single dominant dynamic light source (=sun) and lot's of foliage (alpha masked materials).
(click image for full-res)
-- Edited by mAlkAv!An on Saturday 27th of October 2012 08:54:30 AM