So my original plan was to upload this December a revisit of my nanite vs LOD video going into more detail and doing a more fair test than in the first video.
This gonna have to wait a bit 😅.
Long story short, its taking me more time than anticipated to setup both an optimized LOD scene and an optimized nanite scene. Also I upgraded my GPU from a 2060 to a 3060, meaning I'll have to re-shoot some of my benchmarks.
But the main reason is that I've been messing with the vite fork which I'll highly recommend looking into, and I'd like to make a DDGI vs Lumen comparison before I work on the nanite vs LOD video. I'll post more here and on my twitter as I get things done so you can follow me here.
First attempt at speedtree, very powerful but very confusing, especially with the ways it handles external material/meshes.
This is what I'm working on for the next UE nanite/lod comparison. Again the goal is to have a more fair test environment than in the last video.
Also as stated in the last post, I plan to use voxelize nanite for foliage, from what I tested already it has a few issue but I'll wait for the preview phase to end, maybe they'll be resolved then.
So I've found a bit a free time to work on the next video where I will do a nanite / lods comparison especially with foliage, and for that I really wanted to try the new voxel feature in UE 5.7 preview.
However this version looks like it's a bit buggy and I'm having some very odd behavior especially with shadows and stuff.
Anyway I'll wait for the production ready version to come out to really test things out which may delay the video a bit.
Sneak peak at the upcoming video. So this will be a more fair and deep dived comparison between LODs and Nanite, as my previous video was far in favor of LODs and glossed over many details.
It's my first time making game ready pine tree, for a first iteration they didn't turned out so great and there is room for optimization and visual improvement, maybe I'll give a try to the new unreal tree system will see how it goes.
Sadly I won't be able to work on the video for the next 2 weeks so expect to see this video land somewhere in November or December (December if not everything goes according to my schedule). Or never if I die, who knows.
For the few people that has subscribed to my channel following up the latest vids first a big thank you.
But I have some precision to make about this video which is far from perfect, so I'll just copy paste what I wrote in the pin comment :
Since I posted this video it somehow blew up and with the many comments pointing out mistake I did in this test I have to make some clarification. First this scene use only one mesh, which is far in favor of LOD (In term of memory, In terme of draw calls), second even if I mention LOD popping at the end of the video it would have been nice for me to mentioned it earlier in the video, also about MSAA this scene was kinda an easy win for it and I failed to mentioned many MSAA cons (not working in deferred, not resolving stuff like shader aliasing, texture aliasing compare to most post process AA).
This is why I am working on a follow up to this vid in a more realistic scenario, with more assets and scattered foliage, and redo the test. I cannot guarantee when it will come out, I'm a busy man.
Again this video was just me as an artist exploring my option and posting my findings, DO NOT take this video for absolute truth, best thing to do is to do your own research and see which tools fits your project the best, I was kinda biased towards LOD and forward shading in this vid since my game I'm working on is a static environment with mostly static lighting. Again if your hesitating after seeing this video, do your own research, not everything is black and white.
Antekariel
So my original plan was to upload this December a revisit of my nanite vs LOD video going into more detail and doing a more fair test than in the first video.
This gonna have to wait a bit 😅.
Long story short, its taking me more time than anticipated to setup both an optimized LOD scene and an optimized nanite scene. Also I upgraded my GPU from a 2060 to a 3060, meaning I'll have to re-shoot some of my benchmarks.
But the main reason is that I've been messing with the vite fork which I'll highly recommend looking into, and I'd like to make a DDGI vs Lumen comparison before I work on the nanite vs LOD video. I'll post more here and on my twitter as I get things done so you can follow me here.
See you one day.
5 months ago | [YT] | 12
View 1 reply
Antekariel
First attempt at speedtree, very powerful but very confusing, especially with the ways it handles external material/meshes.
This is what I'm working on for the next UE nanite/lod comparison. Again the goal is to have a more fair test environment than in the last video.
Also as stated in the last post, I plan to use voxelize nanite for foliage, from what I tested already it has a few issue but I'll wait for the preview phase to end, maybe they'll be resolved then.
6 months ago (edited) | [YT] | 10
View 2 replies
Antekariel
So I've found a bit a free time to work on the next video where I will do a nanite / lods comparison especially with foliage, and for that I really wanted to try the new voxel feature in UE 5.7 preview.
However this version looks like it's a bit buggy and I'm having some very odd behavior especially with shadows and stuff.
Anyway I'll wait for the production ready version to come out to really test things out which may delay the video a bit.
See you then.
7 months ago | [YT] | 21
View 0 replies
Antekariel
Sneak peak at the upcoming video. So this will be a more fair and deep dived comparison between LODs and Nanite, as my previous video was far in favor of LODs and glossed over many details.
It's my first time making game ready pine tree, for a first iteration they didn't turned out so great and there is room for optimization and visual improvement, maybe I'll give a try to the new unreal tree system will see how it goes.
Sadly I won't be able to work on the video for the next 2 weeks so expect to see this video land somewhere in November or December (December if not everything goes according to my schedule). Or never if I die, who knows.
Anyway see you then.
7 months ago (edited) | [YT] | 29
View 3 replies
Antekariel
For the few people that has subscribed to my channel following up the latest vids first a big thank you.
But I have some precision to make about this video which is far from perfect, so I'll just copy paste what I wrote in the pin comment :
Since I posted this video it somehow blew up and with the many comments pointing out mistake I did in this test I have to make some clarification. First this scene use only one mesh, which is far in favor of LOD (In term of memory, In terme of draw calls), second even if I mention LOD popping at the end of the video it would have been nice for me to mentioned it earlier in the video, also about MSAA this scene was kinda an easy win for it and I failed to mentioned many MSAA cons (not working in deferred, not resolving stuff like shader aliasing, texture aliasing compare to most post process AA).
This is why I am working on a follow up to this vid in a more realistic scenario, with more assets and scattered foliage, and redo the test. I cannot guarantee when it will come out, I'm a busy man.
Again this video was just me as an artist exploring my option and posting my findings, DO NOT take this video for absolute truth, best thing to do is to do your own research and see which tools fits your project the best, I was kinda biased towards LOD and forward shading in this vid since my game I'm working on is a static environment with mostly static lighting. Again if your hesitating after seeing this video, do your own research, not everything is black and white.
7 months ago (edited) | [YT] | 24
View 4 replies