Frustrated with what's possible with game engines

Win

Just want to commiserate a bit about the limitations of game engine rendering. The workflow for making assets is just grating (retopo, baking). You can't model details individually, especially with vegetation. Real highpoly quality is impossible. I'm so done with this crap.

Kanni3d

dang

poopipe

I'm not sure what the alternative is tbh.
 It's not like you don't have to retopo, bake and worry about being efficient when working on film or TV projects. 

PolyHertz

I'd imagine a voxel engine would be more to your liking. No baking or worrying about how much detail you can add since everythings consistent in that regard. If you haven't already, check out MagicaVoxel and see if you find that type of art enjoyable.

Also for hard surface, there's the face weighted normals workflow which requires no retopology or baking (outside of any decals you don't already have).

And of course there's the upcoming UE5 which will support high density meshs via its Nanite system.

dreamylake

I made a post a few days ago about a new real-time engine that has unique workflow for smooth assets where you import subdivision control mesh and the engine takes care at a pixel level of rendering it smooth. Not sure if this helps with retopo but it sure gives high poly output.

pior

lol

gnoop

Well. On the other hand  the game workflow is stiil having some distant resemblance to an art or at least art related craftsmanship   while  other workflows are  just this:

 

https://youtu.be/DfGs2Y5WJ14?t=18

Eric Chadwick

sorry, not sorry

Win

Eric Chadwick said:

sorry, not sorry

you're a douchebag

Eric Chadwick

Haha, sorry, I meant it as a joke, I don't mean to offend. 

We've all been there. It sucks constantly hitting walls, until, well, it doesn't (at least for a little while). Keep at it, it's worth the effort.

In my opinion the real key to success is to simply get comfortable with failure. Even after more than 20 years of doing CG, that's still a constant for me... learning, failing, and learning from those failures. It's a forever process. If you're not failing, you're not learning, I believe.

Anyhow I hope that helps. This is also a great read in the same ballpark: Motivation is Bull

poopipe

Wow, Eric. 

Thats tough man.  Ive managed 20 ish years in the biz and never failed at anything 😁

Eric Chadwick

https://youtu.be/z_MmCE73bXk

FourtyNights

To be honest here, I'm slowly moving towards offline rendering, but still keeping most of the workflow in the "optimized game asset" way, and using real-time techniques.

Kind of like this:

Realtime in mind -> optimized low poly assets, baking AO and normal maps, hair cards

Offline in mind -> reflections and refractions (mirrors, glass and character's eyes, for example), lighting, shadows, GI etc. Also still creating medium to mid-high polygon count on the assets, like 64 sided cylinders, bevels, semi-dense topo on characters, and so on. Treating almost everything as an important hero asset, with details. No LODs, just LOD0.

Especially characters' eyes without faking depth of the iris with parallax mapping anymore... but really creating convex cornea and concave iris/pupil meshes for the raytracing to handle the realism.

...which comes to the recent version of Marmoset Toolbag 4 with raytracing. Raytracing pretty much everything is what I'm going for, and less faking or using complex tricks to make something to look as good as it's possible in real-time rendering.

The biggest reasons I'm starting to like offline rendering nowadays (I used to be "a huge devoted real-time rendering for games only" fan), is the improved speed to see results faster with RTX acceleration and denoising. And faster rendering in offline in general, as long as you're having a powerful PC, as I have.

Eric Chadwick

If you're making images or movies, no need to work under realtime restrictions.

If you're working on assets for use in actual games in production, realtime methods are still a requirement. Raytraced or not. It has to render at interactive speeds, period. And usually it has to do so across a range of hardware, so LODs.

If you're working on a portfolio piece, don't worry so much about realtime constraints. Visual quality is king! If you want a job in games, it certainly helps to incorporate some realtime production methods. Like reasonable texture sizes, trim sheets, screenshots from a realtime engine, etc. But visual quality is still paramount.

My 2 cents.

FourtyNights

Sure, if I'll be working on an actual video game, I'll use all my game asset workflow skills.

Nowadays, I'm kind of stuck with just portfolio pieces, and I'm eager to use fancier tech to get that amazing final look. Waiting for the day when film and games mash together without a difference in the workflow. The gap in tech between films and games is A LOT smaller than like 10 years ago though.

For me, I'm partially giving up on certain real-time restrictions on characters, and incorporate some offline (aka film) techniques instead, like true raytraced eyes and skin (SSS and translucency) and most likely I'll switch to particle hair as well. But I'm not completely giving up on hair cards though, an useful skill to have.

oglu

Maybe check out Omniverse. Most advanced stuff possible. 
Its in public beta. 

https://www.artstation.com/artwork/KaXZdG

gnoop

Challenges are what makes game art interesting and creative field to work in.   It's where you often have to  invent  rather than learn, while the later is also very important.       Where you are still able  and often have to risk.  With ideas and approaches, notwithstanding  time, deadlines etc.     Well in some projects you can  at least.

  It also tends to turn into robotic pipeline   but still has lacunas where you can find yourself doing something   unique and not easily re-creatable    that  could save you from turning into instantly replaceable workforce .  That with offline  rendering is often as quick as a finger snap with all those armies  of artists from around the globe  that are all perfectly cappable to learn every thing you had learned.    

Eric Chadwick

The latter is not true in practice though. One offs maybe ok here and there but consistent quality is difficult, management overhead is significant, etc. 

fearian

People here might reply that challenge is fun, or that the tech is getting better- basically take it at face value that you have encountered the bleeding edge limitations of what is possible in realtime rendering.

I'm going to tell you that you haven't, and you just flat out don't know enough.

I don't mean you don't know the right tool, or you haven't discovered the right workflow, I mean that I'm assuming from your post that you are not a technical artist or rendering engineer, and that you don't know the ins and outs what is actually possible with a games engine. To put it another way, if I gave you a graphics programmer, a tools developer and a technical artist, and set them to work assisting you, do you think you could achieve whatever it is you are looking to do?

Is the limitation realtime rendering, or you. Your time. Your resources. Your knowledge.

There is a lot of things that can only really be accomplished with a multi-skilled team working together. It's bit of a bitter truth, in games, and working smartly around the limitations of what you can do alone is a soft skill to pick up.