The beta was fun as fuck and the graphics were pretty impressive.
The thing I noticed the most was that while most games try have "muh polygon budget" Battlefront really went wild. I couldn't find any jagged edges.
I think that this will be the future tbh. Hopefully we won't have to be using Normal/Displacement maps for literally anything that needs a little bit of detail.
The game was also beautifully optimized.
Unfortunately, jew EA won't ever release a Frostbite SDK.
And the closest we have is UE4, which is unoptimized as fuck.
>UE4, which is unoptimized as fuck.
Explain this claim.
yep, you're an idiot
It's called photogrammetry m8. People have been doing it for years. Only the static landscape meshes look good because they are actual scanned things and don't really work well with dynamic lights.
The rest of the game, especially the characters looks like the old crysis graphics. Maybe not even that good.
...but it runs at 60fps anon
Bbaked things that don't interact with light run faster. Why is it surprising ?
You continue to shortchange all the hard work the team did. You sicken anon.
photogrammetry requires ridiculously low effort. you take photos from different angles, then the program gives you a mesh with textures on it and decimates it.
>Do any of you think you will but Battlefront just to study it's graphics?
Also you misspelled "buy"
do you think frostbite is "optimized" out of the box or something? you think you could make an optimized game if you had access to frostbite? you don't think they spent some of those $200 million dorrars and years of development time doing that?
you can optimize UE4 to your hearts content bro, go ahead
But the software and special cameras probably cost millions. You also needs tens of thousands of photos, if not more.
You rent them, you don't buy them. Also lasers and point cloud data, not tens of thousands of photos.
i'll just watch youtube vids to study the lighting.
the graphics are nothing special.
Wow, sounds pretty easy. You must have made a few of your own AAA titles the way you describe it. Tell us anon, what bullshit wip are you working on right now?
you realize photogrammetry for professional application requires hundreds of photos per object?
I worked with people who did photogrammetry as their profession. With a decent equipment all you have to to is take shots from different angles with a tripod. The program automatically does the rest of the job to the point of creating the mesh with the texture on it.
You make it sound like someone has to manually sort through them.
of course you manually sort the photos
photogrammetry software doesn't produce good results if the ISO/DOF is varied, which is why you take hundreds of photos and pick the ones that work well together, plus you need to get rid of all the blurry ones because you can't use out of focus photos in the software
also it's not as simple as just pointing an expensive camera at an object, you need advanced lighting setups, you can't have any glaring highlights/shadows or you will get bad results
additionally, the meshes and textures you get from the software are far from production ready, you have to remesh everything and compress all textures to be able to use them for materials
developers don't use photogrammetry to simplify the process of making 3d stuff, it just adds more work, they use it because they need world class, highly realistic and natural looking props
> of course you manually sort the photos
no they don't. using tripods gives you non-blurred sharp photos every time.
> photogrammetry software doesn't produce good results if the ISO/DOF is varied
not touching any buttons sound pretty difficult
> you need advanced lighting setups, you can't have any glaring highlights/shadows or you will get bad results
they don't use any lighting rigs you illiterate faggot. do you think they place any lighting source for entire cliffs and for huge geologic structures and shit ? They shoot everything under natural lighting conditions. Not to mention stones and snowy mountains don't reflect anything under overcast weather.
> you have to remesh everything and compress all textures to be able to use them for materials
wow...pressing remesh button is a real challenge i'll give you that. and compressing textures using photoshop ? now thats a rare skill my friend
They use photogrammetry and its piss easy. Even amateurs do it successfully because it requires no talent or artistic impression or years of experience in sculpting with zbrush. You're just trying really hard to find baseless excuses to suck their dick for fuck knows why.
Why don't you all just watch what Epic had to do to produce their Kite Demo assets?
>7-45GB of photo data per asset
>Requires massive computation to turn into high poly model
>8-20m polys, 16K texture
>This then has to have lighting removed from it to make them game ready
>Reduce poly count of asset, make LOD meshes
>They shoot everything under natural lighting conditions
if you don't think they use lighting setups when photographing then you have zero idea what you're talking about
look up any documentation on large scale photogrammetry and you'll find tons of info on how lighting is set up for outdoor scenes
>do you think they place any lighting source for entire cliffs and for huge geologic structures and shit
they use them for smaller objects as well or you would obviously get poor lighting 100% of the time
>no they don't. using tripods gives you non-blurred sharp photos every time.
yes, they do
if you're gonna spend months on making props for a game you obviously take the time to go through the photos you use to produce them
>Even amateurs do it successfully
no, they don't, amateur photogrammetry 99% of the time produces poor results unless it's very simple objects in which case it's irrelevant
You're massively overestimating the effort required for photography. Light isn't so fickle that you need studio lighting setups for literally every professional photograph, and photogrammetry programs aren't so poorly-functioning that they need 100% fidelity between shots. So long as the shots aren't ~30 minutes apart, on a clear day you're going to get lighting conditions so similar that variation wouldn't be noticeable.
You need to double polarize the subject. Having done a fair amount of photogramatry, to get good results you do have to take time. Lots of pictures, lots of pictures with markers.
The mesh clean up is easy, but its easy to be on set and think you've taken everything to find out later your missing some shots. Then you have to mess around and it isnt a simple retopo job.
Also even with high resolution images you'll usually have to do some clean up sculpting because you get blobs around the mesh.
I'm with the other anon. Get off the fucking Battlefront dick and realize that these graphics aren't that hard to implement into game these days. Especially considering it's debatable that these are actually amazing graphics.
>plan to make my own game engine using opensubdiv ptex and vulkan while im going to university under the hope that by the time i'm done uni it can run on consoles easy
am i going to regret this /v/?
Yeah you're gonna regret that
There is really nothing special to study
OpenSubdiv isn't meant for the kind of real-time performance that games need... It's good enough for viewport where it will be choppy now and then, but when run alongside all the other shit that needs to be processed in a game, it will be too much. Just use the built-in tessellation and displacement features that are part of OpenGL, DX11 and I'd hope Vulkan.
PTEX is also not meant for real-time application bruh, it takes more processing power than traditional UV mapping, it's simply better for production because at the cost of performance, you can achieve better texel density, helping you save on memory usage by not needing to use ridiculously high resolution textures. Production scenes use ridiculous amounts of memory, so every little bit helps in achieving higher fidelity. Fidelity > Performance when it comes to film.
Instead of PTEX, you should be implementing Megatexturing in your engine.
The fact you understand none of this is a pretty good signal you will fail though.