Speed
admin
Hi guys, I'm a bit worried about my speed in completing assets/objects, comparatively the high poly side doesn't seem to be a problem but the rest is long winded. For example this generator has taken around a week and a half everyday for a few hours, I'm almost finished baking out all the bits and so haven't touched on diffuse at all:\
This has probably been asked before and is a difficult question to answer but it would be nice to know from you who work in the industry: How long are you given to make a prop of this size that would appear once in a game? Also something like this that would appear in every level?
Thanks
This has probably been asked before and is a difficult question to answer but it would be nice to know from you who work in the industry: How long are you given to make a prop of this size that would appear once in a game? Also something like this that would appear in every level?
Thanks
Replies
may i ask what program you are using? applying custom hotkeys and using various plugins, knowing and making use of all given tools will increase your speed alot. in case you haven't done that yet.
but in general, speed comes with routine/experience. already knowing not having to think about getting certain parts done. making less mistakes. stuff like that...
so i wouldn't be all too worried about it.
Once the scale is determined, and you get into the nitty gritty of modeling stuff like this out, the biggest time sink can be overall workflow issues, or whether or not you are modeling "smartly" when creating your shapes.
Do you have scripts, plugins, hotkeys, and extra stuff to help you with your workflow? Do you use both hands to model? Are you using dual monitors or only one? (Makes a world of difference) Instances? How is your baking process?
Are you wasting a lot of time modeling stuff for normal maps that you could do just as easily with Photoshop? (Small bolts, letters, insets, etc.) It really comes down to your workflow, identifying what is taking you the longest to do and figuring out how to improve on it.
Personally, the high poly for this would probably take me about 3-4 hours or so, if that. It's mostly all boxes and a few custom shapes. I wouldn't even bother with a lot of stuff to do a high poly for. (See above) I could probably have the whole thing textured and in Engine before the end of Day 2 and not really break a sweat. I'm not trying to brag, I have been doing this for awhile now. It comes from repetition and knowing the tools.
Above all, aside from having useful tools at your disposal, repetition is the key to speed. My unwrapping is pretty fast because I use hotkeys, use scripts, and know what is going to result in a good bake / texture job.
How long am I given for something like this? At High Moon we are either given tasks or just "know" what are tasks are, depending on whether you are on an environment artist or a prop artist. You determine how long you think it will take and estimate those tasks to your producers and leads. If you go over, or are slow, you are only hurting others around you and the team as a whole is let down, so we work hard to estimate correctly and bust our hump to get it in a reasonable amount of time.
Personally I would probably estimate this to take me about 2 days to complete.
I really feel silly looking back at my work in games and complaining we didn't get enough time:)
I figure that asset could be done in a day easy
No point in spending so much time on whats essentially the kind of asset a player will look at for maybe 2 seconds in a game (that doesn't mean it shouldn't be good tho! Just not super-perfect-hipoly-normal-map-everything-unique good).
A bunch of boxes to block it out, bevel some edges, use a few splines extracted from primitives to do the fancy tube/pipe work, instance a bunch of bolts, vents and switches around the place and the thing's done.
With decent reference the highpoly for that shouldn't take more than a few hours to an experienced highpoly modeller.
An hour or two more for the lowpoly, unwrap and bake, then a few more hours for the texture. 2 working days (assuming 6 hours of work gets done per day) would be a safe bet for finishing this to a good standard in a working environment, assuming it's intended for an FPS or close-in third person game.
Any other game type and I probably wouldn't bother doing a highpoly, since it'd probably be seen further away or at higher speeds.
Really you should either do a HP, or not do a HP, doing like.... half a hp model? And doing the other half of the detailing in photoshop to me seems unquestionably slower. If you're taking the time to bake anything, take the few more minutes to finish the HP model, instead of adding another phase to your developement.
example of easier in PS: a large scratch/gouge in the metal. I'd crazy bump rather than sculpt that if time was an issue.
Do some zbrush, do some xnormals, do some high poly modeling, run the gamut and keep an open mind. It all falls together with repetition. Think of what works well quickly, and think of what looks best even if it takes a long time, find a balance that works for each model.
Planning, Modeling, UVing, Texturing all get better to various degrees with each model you make. Look at the big picture and understand that each is vital to the quality of the end peice. Don't expect to be epic overnight and just keep on truckin'.
The harder you fail with a model the more you learn.
when at the beginning you're doing something new, or a workprocess you're not quite used to, it'll take some time, but later on, if you've done that kind of thing a couple of times later, the workflow will be familiar, and it'll go fast.
ps. the time spend quickly modelling things on the highpoly will save time cutpasting different maps and normalmaps in photoshop trying to get them to match :P
That's more or less what I was saying as well. If it's just a quick nub for a bolt, sure, modeling it is gonna work better. Modeling HP letters is a damn pain though
One thing that I don't think you've realised is that you don't actually have one model there, you have several potential models.
I advise you take a look at what you have as though it is lego, and try building 3 or 4 new models using the parts that you already have. This may require some extra texture work, and new AO, but I'm sure that if you succeed with building 4 models out of what you have, then you will have increased you're productivity, and potentially cut a single 2 week task, into 4, 3.5 day tasks.
If you keep all the parts, or sections of assets you make in seperate files (I'm talking about all the models you make in the future) then you are essentially building a pick and mix for you to delve into whenever you need something. The more you build up, the quicker you will be able to produce assets. I realise that in many cases changes will needed to be made to pretty much every piece, in order to make them into one decent model. But when it comes down to mechanical objects, or something like stonework and masonary, then having a bunch of generic assets to hand can be a real time saver.
-Adam
the fugly mesh:D
I even wrote up a quick tutorial on this subject.
I'm almost tempted to try modelling an asset like this today just to see how fast I can do it (I've got the day off
Even if the model has to be true to a reference, the very nature of game screen resolutions, and realtime lighting, make bolder models look much better. And as a handy consequence, such bold models are faster to model than blueprint-accurate renditions.
The best examples of such bold designs are on Paul Richards/Autodestructs folio. 45 degrees is where it's at!
[edit] just wanted to quickly illustrate my point :
Bigger details, faster to model!
But what you are saying is very true. A lot of 3d artists tend to focus way too much at all the small details, zooming in superclose to the model when working on it. While all that really matters is how the model will look from the distance the player will actually see it. You'll often need to exaggerate details to make them nice and readable. For example big bevels will give you nice highlights on edges, small bevels might be more realistic on your hipoly but once baked down (if it's picked up at all) it will only end up being a bunch of ugly flickering pixels on the edge.
yeah true
lefix - 3ds max is my core program, not versed in any other 3d software apart from Zbrush really. Few vague memories of Milkshape and Blender but that was years ago!
You mentioned hotkeys, have a few setup for common tasks, "u" for unwrap and "s" for stitch verts etc. but don't use any plugins or scripts. Max is running no frills as per the installation though this gets me thinking what tools would be publicly available and really beneficial to shave off some serious production time.
Ott - A guesstimate of the scale of this prop in comparison to a character is around 1/6th size. Not really sure what perspective I was thinking it would be played in, maybe fps.
Again no extra's used in max, yes to two handed modelling but no to dual monitors
Yeh, everything was modelled down to the nuts and bolts, even bundles of high poly wires. Used one modification and then mostly duplication for a new nut type and pasted them around the model, so it's about 50% unique.
The baking process is really touch and go, taking up lots of time and therefore my least favourite part. With big/complex models, small chunks are done, then maps arranged after the uv's have been moved around a few times to incorporate new elements. My workflow sucks, I know!
Thanks for the insight into your job at High Moon, this is my fear of holding the team up, that is, if/when I can get a foot in the door at a game studio.
Proxzee - phew, it's not just me
Ruz - Completing this whole asset to aaa standard in one day is something that feels unimaginable to me atm. You must feel egomaniacal after pumping out stuff that quickly, kudos.
ae. - your super fast too then (not that anyone repying here is slow) but are you talking about repetition or the undiscovered?
Canadian Ink - That feels like relief, I've never been paid for any work I've done, it was all for pleasure and the community. Hopefully when that happens, the team will introduce me to time saving aspects in a proper production pipeline.
Peris - Presume you mean modelling the low poly first, then example, a high poly nut. Baking and duplicating it before modelling the rest of the high res. That's something I had considered but didn't like the sound of, if it saves time, cool.
MoP - That's exactly the workflow used, the problems started when realising that this is 2 seperate meshes at the least. At that point I started concentrating on the generator, not the whole air conditioner. Unfortunately only a couple of parts from the unfinished air conditioner were able to be re-used in the gennie.
This reference was the only one of it's kind I liked/found. The end bit where the gennie is now is pretty much unreadable so I used multiple references to make an oldish looking gennie from what were mostly very modern ones.
aesir - Sounds like 2 days is the average time for industry peeps then, good to know some limits
crazyfingers - It doesn't seem to be repetition for me, every object has it's own unique features so baking is guesswork and very time consuming. Not having worked in the industry is maybe making my experience biased though.
You recommend trying to modify workflow to not spending most of the time in one app/area, currently trying to live up to that but it's damn hard to break those habits. It's like only when a certain app is required will I use it, is that bad practise or practise that only the veteran should adhear to?
From looking back things are improving slightly without being bigheaded, it feels good but also awkward to know you failed so many times before on not doing you best. Not sure if that's pessimism, realism or a bit of egotism speaking but it feels like I'm kicking my own ass either way!
Thanks for the support
TeriyakiStyle - You hit on a few things there that I started doing in the last year or so, keeping a fairly vast texture library and only very recently keeping a few high poly parts. This brings me to another question, a bit off topic but one I want to ask : Say your familiar with a 3d artist and then see some work of theirs after some time and recognise it saying "o, that's them" is it because of the cloned parts they re-use in models or would you say it is more an overall style?
Thanks for the tips
eld - Thanks. Do you think it's a good idea making and keeping high and low poly parts, then if an object is a different scale both low and high res could be scaled down to suit the current project?
Bad Spleen - Actually I did know about having multiple meshes here, before modelling even a blockout, the model was broken apart in the mind. Just ran into a snag when realising the reference was unique and part of it was obscured badly! Nice idea about saving modular stonework, have yet to do any real stone modelling, soon though.
Cheers.
Disco Stu - Inset text. Are you talking about using floaters or to boolean away the surface of the mesh it's going to sit on? Which is the best method and what difference will it mean to baking?
pior - Know what you mean! For instance : 2 hi-poly bundles of wires were made out of splines, simple 5 sided cylinders surrounded them. Out of all the bakes it was definately the worst, infact it was totally pointless and kind of made me angry at the conclusion.
The larger forms as you say have recently been noticed on the modelling side and not just in texturing as previously. Suppose it should have been one before the other, but I'm glad you mentioned this as it seems to reflect in the end product quite significantly and is most obvious from your paintover.
Thankyou.
rasmus - The texture size for the gennie is 2048px squared, uv's are well optimized and planning 3x 512 decals for some badges and the voltmeter readout. Been thinking this will be one of two objects out of the whole thing - although, maybe it should be split up more, if not for texture clarity, for useability?
About the hero piece, that is what I was doing! It is for portfolio so would you say an employer would prefer something not so polished/defined made in a much shorter time to this extravagance?
Thanks for the tips and support.
You sure? It looks like a pretty damn big thing. Halfway up it's table height, and those switches and buttons on the front would probably be around chest/shoulder height by the looks of things.
It's certainly longer than a person lying down, I'd say, probably about 2 metres long and 1.5 metres high.
I'\m guilty of doing the same thing as you in that sense in many occassions. Working from home makes me slow, but when I work at the office with no distractions and a huge window on my back I churn stuff out like water.
It's easy to say it would only take a day or two or three or whatever, but things always seem to take longer than expected. You run into things, especially doing Highpoly or baking that slow u down. It also depends on your studios pipeline and work distractions and just so many other things. Max crashing, computer crashing, bathroom breaks, ect.