Hey there, this question have been on my mind for quite some time.
Why don't more people use CryEngine/UDK as their 'testing' and presentation programs for video game assets creation?
Isn't Marmoset, CryEngine, and UDK are basicly all physically accurate and "unbiased" rendering engines? Why does 'everyone' using Marmoset Toolbag? Isn't it BETTER to show off your model in an actual game engine?
Thanks!
Replies
Rendering in CryEngine or UE4 (not UDK, UDK is not physically based) are both solid choices if you already have a good working knowledge of the engines, but you need to remember that they are first and foremost full game engines. This means that they come with a lot of "clutter" if all you're trying to do is quickly render out a portfolio piece.
This can come in the form of large initial download sizes (bad for people on limited bandwith plans or slow connections), or it can come in the form of additional/unwanted features, and more commonly it comes in the form of overly complex functionality... Remember that the goal is to quickly iterate and render your piece, so you don't NEED all of the failsafes and behind the scenes measures used in full game engines to make sure assets are game ready. But those things are inherently there in those engines, they have to be.
Toolbag in contrast is lightweight in terms of filesize, and also featureset. We're doing our best to make sure that the preview/review/render process is as quick and painless as possible for artists, while maximizing visual fidelity. And that is why i used Toolbag before i joined the team, and why i believe other people use Toolbag over fully fledged game engines when it comes to portfolio work.
There are also studios using Toolbag in their production pipeline, by either altering Toolbags shaders to match their in-engine shaders, or vice versa. Because they believe that Toolbag is the quickest/easiest method of previewing work for their artists during live production.
Render engine is probably much more powerful in Marmoset over game engines because it does not have to handle as much at the same time as a game engine if that makes sense.
Finally someone who thinks differently. So does it cost you in image quality and you just don't mind it, or do you manage to make it look as good as Marmoset would have?
Below is a good example:
http://www.aldenfilioncg.com/
Yes, BUT ... Complete environment shots from games can be problematic. Art directors here have complained about that as unless it's spelled out, they don't know what you did. Which parts are yours? The entire thing? Lighting, FX, everything? That's not usually the case.
the main benefit to us is that it lets us set up consistent looking renders of our work whereas when you deal with a game engine, it's features change all the time and you're dependant on somebody to set up and maintain a level with suitable lighting for you and may even have to pull stunts to get a video out, have perspective distortion to deal with, etc.
stuff is of course cross-checked in engine to make sure shaders match and we're not using features the game will never be able to run in the first place.
- Hard to get transparent background with UE4 shots. (I use one of debug view modes as a mask)
- High resolution screenshot renders with low-res mipmaps.
+ Custom shaders. If you're into shader magic, it's the only choice
+ Quite easy to bake lighting (with indirect bounces), so it's good for enviro fragments.
EDIT- I just remembered, if you're shooting for anything requiring Anisotropic highlights such as hair, UE4 is not for you, and Marmoset 2 may be the renderer that fits your project best. Epic has not implemented anisotropic shading yet due to the high render cost in a deferred-lighting engine. Than being said, if you look at this as a challenge rather than a dealbreaker, more power to you.
When choosing what to render in, you should ask yourself what you're trying to say with your project. Is it simply a means to an end? Do you wish to demonstrate proficiency with a particular engine? Are there any technical requirements you have for the project, such as custom shaders or modular assets? What's your time frame? Who or what is the render for?
It may be different elsewhere, but one of the questions I've been asked frequently for work is: "Are you familiar with <insert engine here>?", and I've heard a fair bit of complaints regarding artists who have good renders via Toolbag, but lack the technical skills or workflow knowledge to actually get the assets in-engine. This may only be a concern if you're headed to a smaller studio or an art house where you'll wear more hats than can fit in a decent-sized haberdashery, but it's something to consider.
I have yet to see someone mention for example shadow issues pending on the card type and drivers you might be using with a specific engine (shadows in real-time don't work in Autodesk products on AMD 8xxx series cards and meshes don't show in the Source engines with specific drivers, or mipping issues with certain transparency parts, etc), things that actually do matter because they impact your asset in a negative way.
Everything else is personal taste, why does it matter if it's an engine or a offline render? VRay for example still doesn't have GGX Specularity while UE4 does, BUT reflections are impacted in vice-versa. Toolbag is better vs both, but the SSS features will often throw Radiosity like speckles on your mesh if you set it too high in light pass option vs. offline or less physically correct SSS methods.
You can take a look here for a few of the mentioned issues: http://www.polycount.com/forum/showthread.php?t=136390&page=5
That's silly. If it's not a public engine or one that uses COMMON importation systems in place with reference files like UDK, (ei: reinventing the wheel on their engine instead) then I don't think they have the right to complain, especially if their engine is something like GameBryo that dumps entire lists on your face and can be daunting for any artist that doesn't have somewhat technical centric background on text driven searches (outside of bugs that make you lose work).
Also, RAW skills from engine to engine don't translate that well if they are radically different unless the artist has a blueprint they are allowed to copy and paste and use as a base, (like using same default print for a piece of armor, and replace that with a different mesh and texture call with a unique item list and stat, ei GameBryo) and which point, I have to ask, what exactly is this 'engine technical skill' they are talking about (everyone can Ctrl+C and V blueprints)? It's the equivalent of telling an immigrant in Canada to go and work at Tim Horton's to gain 'Canadian work experience'.
Also, lastly, if the engine is closed off/limited to any type of internal modifications for rendering materials, you're kinda limited to what you can do. If you wanted to create Avatar: LoK like model and toon material, I'm pretty sure your AVERAGE artist wouldn't be able to do that for next-gen engines like UE4 or CE3 (even Toolbag removed those options in V2 last I checked), so you're kind left with Offline renderer's or engines which allow for large scale material types or modifications (worse comes to worse, you have to buy said shaders as extension) or grease your elbows and get to work on custom stuff.
Way too many variables to make a one point comment.
zearou, your questions should include:
- Which solution shows my skills best?
- Does the solution not affect my work negatively (what Ace-Angel said), because of reasons that won't be my normal job?
- What do I want to show:
- my model and texture (pretty and quickly, because I have ton of them)
or
- my additional skills (FX, shader tricks, engine knowledge)? (Der Hollander's point)
- Is it an environment? ;>
If I could +1 a PC post...