• What are you working on? v67 - March 2017
    3,527 replies, posted
Now with animation: https://drluke.space/pyree/ https://www.youtube.com/watch?v=cRdYjTGg_no
Can someone help me set this up to build in Eclipse? I haven't used java in like a year https://github.com/perfectsense/brightspot-cms
Reminds me of the old Interplay logo http://c-6rtwjumjzx7877x24anlsjyyjx2ebnpnfx2esthttpnjx2esjy.g00.wikia.com/g00/3_c-6kfqqtzy.bnpnf.htr_/c-6RTWJUMJZX77x24myyux78x3ax2fx2fanlsjyyj.bnpnf.sthttpnj.sjyx2fkfqqtzyx2fnrfljx78x2f0x2f04x2fNsyjwuqfd_Jsyjwyfnsrjsy_qtlt.uslx2fwjanx78ntsx2fqfyjx78yx2fx78hfqj-yt-bniym-itbsx2f195x3fhgx3d75625858630651x26n65h.rfwpx3dnrflj_$/$/$/$/$/$/$/$/$/$
https://files.facepunch.com/forum/upload/1857/93150799-37cf-4e7c-96ef-9a962dbe2025/metro_looping_500px.gif Final push before GDC
Did another thingy (they come up random, so you might have to refresh several time to get this one): https://drluke.space/pyree https://www.youtube.com/watch?v=58hEer8c4e4
Looks like I made a small typo in the original source file in AddMesh- The typo was not what produced the error, and was made when editing out comments for clarity before posting. The [I]actual[/I] code being used to render is here; [url]https://pastebin.com/QMfhY5gd[/url]
Your website's certificate expired fyi
I knowwwww, for some reason the letsencrypt certbot can't read the challenge file from the webroot and I couldn't be bothered to fix it over the last few days.
I'm like a week late, but just to note that if GL_ARB_bindless_texture is supported you don't need to do any of the texture array stuff. Sadly it's not a core extension and it's unsupported on macOS and Intel, but modern AMD and NVIDIA cards implement it. With GL_ARB_bindless_texture you can pretty much get the draw calls down to a single draw call per shader (as long as your material and object data fits in your UBOs – if not, you'll need to break batch every time you switch UBOs [although you could use SSBOs to get around the size limit, but it might be slower]).
Started re doing all my rocket parts models now that I finished with the resources, here on the left is the old Combustion chamber model ( the part of the rocket engine where chemical mix and combustion happens) and on the right is the new model with real dimensions but also a detailed regenerative cooling pattern on the surface. https://cdn.discordapp.com/attachments/369257830882082827/422899969595408404/2018-03-12_20-22-26.png I based my model/textures on these images published by SpaceX since I plan to model the merlin engine completely https://files.facepunch.com/forum/upload/1287/eb955b4f-f9a2-4dad-b283-1f22dec942f3/Merlin1C_Chamber.jpg
That terrible feeling when you've finished all your work for the day and now have nothing to do I want to fucking die
Any cmake pros or people who know modern cmake? I'd love to ask some questions on a rolling basis.
Added a simple navigation mesh thingy to my game. The little red cube is a gizmo indicating that the spot is occupied. There's a hotbar with a 'raise corpse' ability and a 'sacrifice' ability. The player is ranged, the others are melee. Soooo much to do still. https://i.imgur.com/fGuLCcA.gif
Not him, but I do like this approach I have a single ssbo responsible for holding an array of 64-bit texture handles, where each element in the array represents a material and is a texture array (layer 0 albedo, layer 1 normal, etc)
That's clever. I didn't even think of using texture arrays to store the normal maps and such. I have them as separate uniforms right now.
Well, I gave up on indirect rendering because it looks like it really is not my fault. Indexed rendering works fine, however. So do compute shaders! https://cdn.discordapp.com/attachments/263886545495261184/423326255518056448/Snow_volume.gif
Unfortunately no such thing exists in Vulkan, but I don't mind the current setup since binding calls in Vulkan are already MUCH fewer and further between than in OpenGL. Seriously, everytime I see OpenGL code the binding stuff and how commands are "recorded" makes me feel ill I've got my shader generator pretty much working: https://gist.github.com/fuchstraumer/9802f2fc580be0c0f2dcda5c4f373f31 That's generated from a file that has the same main chunk, but all that prefix'd stuff for descriptor objects + the included structures/free functions is included with these few lines: #include "Structures.glsl" #include "Functions.glsl" #pragma USE_RESOURCES SORT_RESOURCES #pragma USE_RESOURCES MORTON_RESOURCES #pragma USE_RESOURCES VOLUMETRIC_FORWARD_LIGHTS $SPC const uint BuildStage = 0; Resources are "blocks" of Vulkan descriptor objects declared in other files, so in this case the file that is broken into chunks is this. It lets me separate resources by usage, and avoids requiring using all the resources in every stage. Example resource file. Resources are unfortunately currently prefixed with $ since it's an easy thing to regex filter for. Here, "$SPC" is a specialization constant which is somewhere between a runtime constant and a compile-time constant - it can be modified at pipeline creation time, so here I'll have two separate pipelines using the same shader but I have a simple "toggle" to decide which particular variant of the BVH building algorithm to run. Usually, drivers will optimize the unused branch out, instead of actually testing the logic. I will say that I'm glad I finally got this running. it's not the most exciting work, but it beats the everloving fuck out of copy and pasting descriptors + structure declarations + free functions between all my shaders. And that's before I consider the convenience of having it handle the (set, binding) stuff for me, too! Now I need to update/test my Shader processor/compiler, and in total I should have a much simpler and easier-to-use shader ecosystem that provides me useful tools and options from the creation to usage stages.
https://files.facepunch.com/forum/upload/111926/08935930-03cc-4595-8549-f7d07d0ee078/image.png
After pushing some more optimizations, the max particle count is now significantly higher. https://files.facepunch.com/forum/upload/111926/4cb63e89-d5fa-4d10-9eea-5606eb93dec7/image.png
What exactly am I seeing? Are you calculating the particle positions with compute shaders and vector fields?
https://files.facepunch.com/forum/upload/1857/57f0d097-5cdf-465c-a126-8152ba6b9390/mysterious.mp4 Making good pigeon AI
Mulder before he got his job
Still working on my portfolio to swivel into a Junior Developer job somewhere! I made a chat app using websockets that works on local networks. Signed up for AWS to try and get it working over the internet but I got confused and scared of being charged a million pounds so I closed my account and left it like this https://files.facepunch.com/forum/upload/133252/18a2d19b-f2e8-4e0b-b848-cbda07e617ba/notreallyabot.png
You should .gitignore node_modules. Also, you can get a vps for like $5 on digitalocean and for €2.99 on scaleway, I would recommend those instead of aws if you're on a budget and just want to try stuff out.
that looks a little something like what happens when the model normals point in the wrong direction so check your model normals and your normal mapping calculations i guess?
Ah, thanks, I'm still getting to grips with NPM and best practices. I'll defs check out those VPS providers, I am indeed broke and very much looking to just experiment. Cheers!
Cross-posting this from the web WAYWO in the hopes of getting more responses:
Just do resume website, if companies can't appreciate the effort, I'd say fuck that company.
This is what I was thinking as well- Sadly, looks like that's not the case.
Imagine your typical hipster pseudo-Nineties/early new millennium 3D game engine, but this time it's in D, for extra pabst points. Link for the unaware.
Sorry, you need to Log In to post a reply to this thread.