• Advertisement

JoeJ

Member
  • Content count

    897
  • Joined

  • Last visited

Community Reputation

2703 Excellent

3 Followers

About JoeJ

  • Rank
    Advanced Member

Personal Information

  • Interests
    Art
    Programming
  1. Voxelization cracks

    Yes. Maybe some exponential mapping for Y, or just a half float. (But i'm still using float4 myself and have not tried anything of this yet...) In OpenGL culling only depends on winding order, IIRC
  2. Voxelization cracks

    About the encoding, i think it would be better to use more bits for luminance. Something like YUV for example, 16 bits for brightness and 2*8 for color: http://softpixel.com/~cwright/programming/colorspace/yuv/ This way you would have enough range to support area lights made of voxels.
  3. Voxelization cracks

    Seems the visualization itself is wrong, because some voxels don't have 6 sides but less. Wrong volume data alone can't cause this. (Look at the single black voxel at the base of a column, near the center of the left half of the image.)
  4. 2D what am I doing wrong? (concept art 2d)

    What is wrong is, since this is assembled from multiple photos(?), the lighting is inconsistent - it comes from random directions. (Mainly visible an the pyramid.) You could fix this by repainting the image using the current as a template (usecase portfolio to get hired), or you and your team may decide to accept it this way to save time (use case already working on a game). You could also make a compromise: Mirror and brighten the pyramid, mirror the burried sphinx statue so light comes from top. Would make a big difference i guess. there are also tools to change the lighting for a photograph semiautomatic. I don't know if Photoshop has this in already.
  5. For example we want to the user to be able to lerp between male and female by a slider, you can do this by modeling both from the same topology. Assume we already have a male base model. Move all vertices so it looks female, but don't add or remove vertices. For each vertex store the difference to the original male position. (we call this a relative 'morph shape', or 'blend shape'). In the game the character then is made by male_vertex + female_offset * gender_slider. You can also handle details this way, e.g. make just the nose large (so most vertex differences are zero). You can combine all options simply by: male_vertex + female_offset * gender_slider + nose_offset * nose slider... The second option to consider is to affect skeleton bones, e.g. Male: wide shoulders, narrow pelvis, Female: narrow shoulders, wide pelvis. You can utilize scale and sheer as well for skinny / fat and so forth. But with caution, as altering skeletons also affects its animations.
  6. Ray tracing 60 FPS on Tablet.

    Missing realtime global illumination is the main reason games don't look real, so any technique that solves this is welcome. Pathtracing can produce realistic results, and the simplicity of the algorithm makes it very attractive because it can handle all optical phenomena with one approach. We are just so used to the thought pathtracing is too expensive that we tend to act critical, but the revolution will happen. Allthough probably in small steps and in combination with other promising techniques (that's what i think - others think pathtracing will just replace current game graphics pretty soon). Raytracing is already used in games, e.g. accurate shadows in the Division. Voxel cone tracing (Children Of Tomorrow) is also a form of ray tracing and gives soft and sharp reflections at limited accuracy, so almost complete GI. (Infinite bounces are missing but theoretically possible.) Pathtracing needs a lot more rays. Not just one (or a few) like for point light shadow or a perfect mirror reflection. It needs 100s or 1000s to capture the full enviroment. That's out of reach, but denoising can help. It borrows rays from nearby locations and from cached previous state. Naive example: 16*16 pixel neighbourhood * 16 stored frames = 4096 rays, so enough. This is why we will change our minds about path tracing. (The downside is again inaccuracy, for instance sharp reflections get blurry.)
  7. Ray tracing 60 FPS on Tablet.

    I did a close look on the webpage. They compare their realtime tracer with an offline renderer and claim a huge speedup. (The offline renderer is made to handle extremely complex scenes AND global illumination, they show neither of that.) They claim a new solution to the problem of building acceleration structures. But i don't need to rebuild the acceleration structure for that dino - i only need to refit it which is a fast operation. (It's common misbelief building acceleration structures would be the main problem of raytracing. The real problem is random memory access and caching.)
  8. Ray tracing 60 FPS on Tablet.

    Note that the dinosaur shows just simple lighting, no global illumination. Thus 60 fps are not impressive but expected. AFAIK, Brigade is planned to be released for Unity this year (don't know if for free, but guess so). OTOY works on multiple path tracing renderers, for both offline and realtime, Brigade is targeting realtime. With recent advances about denoising, path tracing is very close to realtime on high end GPUs, see e.g. this work with GI: Many people work on this actually, many will just implement on their own when time is ready. Hard to tell if AAA studios will use middleware or do it themselves (i guess the latter).
  9. Seems you're quite young and don't know the classics? Here's some more for your list: Amnesia of course, which successfully started the Indie Horror genre. Outlast (the first one): Jump scares, very terrifying and maybe the older Call of Cthulhu Dark Corners of Earth: Adventure / Shooter oh, and Dead Space / Alien Isolation (both more Shooters than Horror)
  10. You can build a motorcycle graph: Find corner vertices. From each of them and at the same time you start motorcycles driving constant speed straight in u and v directions. They leave a trace behind, just like in the movie 'Tron', or in the old mobile game Snake. If a mototcycle hits a trace of another, it stops, the other keeps driving. Finally all traces form coarse rectangular regions exactly like in your pictures.
  11. Trickjumping in games

    Whooo! It works!
  12. Trickjumping in games

    Ok, tried it again using your instructions... but nope. Seems i'm not that everyone
  13. Trickjumping in games

    I played Q3 LAN for years with friends, later Q3 Arena for some time. I love the physics and the flow of the game. But those trick jumping skills never made it up to me. None of them. I tried this quite a few times, looking some video... but i don't get it. So what i think is: This stuff is too unaccessible to the average player, badly introduced and too hard to learn. Probably because lots of it was not intended by the developers AFAIK. But this could be improved and reused in actual games.
  14. Hard to say without knowing more about your motivation for the maze. However, i say i like exploration and narrow enviroments, but i also say i don't like mazes... although both can be similar. So i'll point out what i don't like about mazes as i imagine them in a game: Repetitive, so boring (even if you stress me with monsters, which makes it just worse.). Not knowing how large the maze is, so not knowing if there is a chance to get through at all. Missing motivation to get through at all, eventually what comes next is just another maze. You need to prevent this thoughts to come up. Eventually the story prepares the player to the maze and makes it interesting, so thinking: Ha - finally i am in this mysterious maze there have been so much talk about, now see what happens here... A map of the maze would be a too trivial solution - if there is a map, then why a maze at all? A whole game about multiple mazes in 3D... that's no good idea i guess. Spiders... although such primal fears are not that original. In general i'm hard to scare and to be honest i do not understand well why some movies scare me but others do not... it's black art to me, something psychological... can't give a good answer.
  15. Penumbra (predecessor of Amnesia) I liked this game instantly because it has that realistic physics i always wanted too see. This way the world feels both real and interactive. The game is still the best in this regard. After playing for a while, it really managed to frighten me. Other games like Resident Evil or Silent Hill did not work for me, so i found it surprising a video game can be that scary. First person is a must, believable world also helps here. They did not use jump scares. It is also very good at story telling, mainly by reading notes you can find in the enviroment. The game constructs the story early on and does not throw you into something without background or reason. Claustrophobic and dark without much action and room for exploration. Enemies you can't kill. Not in 3D. People have very different sense for navigation, mine is bad, so maze could be a show stopper for me. If you want it, make sure the enviroment does not look the same everywhere so people get lost. Yes, absolutely. Puzzle / Exploration works better for horror games than shooting monsters i think.
  • Advertisement