Our experience with Blender in a small gamedev studio.

General discussion about the development of the open source Blender

Moderators: jesterKing, stiv

Post Reply
goruka
Posts: 0
Joined: Mon Jun 09, 2008 6:06 pm

Our experience with Blender in a small gamedev studio.

Post by goruka » Mon Jun 09, 2008 6:49 pm

Hi! I work at a small game development studio, we make simple games for handhelds and consoles and evaluated Blender for production, since not only the cost factor is significant, but the alleged productivity of the modelling tools.

First of all, let me say I believe Blender i s an awesome product and while we believe "It's not there yet", it has shown so much potential that we hope we can use it in the future.
However, I will try to give as much feedback as I can from our experience in the hopes that the issues presented will be improved in the future.

To begin, the modelling tools are excellent. Our arists learnt and modelled objects with a lot of ease, arguing that almost no clicks are needed and that everything is in the right place. I was worried that the seemingly complex UI was going to be a stopper, but it ended up being very productive instead. Seems the right polygon complexity can be achieved easily.

The material editing is where problems began. The material editor seemed to be very complex and, unlike the modifier stack. it's difficult to find the where "basic" parameters are. Tools like 3DS MAX allows for material models wit different degrees of complexity.
Since this is 3D for games to be used in real-time, this shouldn't be a big issue anyway, but the issue further became more complex with the UV/Texturing.

From what i've seen, UV-mapping seems to be the weakest part of Blender. I don't mean the UV editor itself, which is probably more advanced than the one in other apps.. as creating UV maps is nice and intuitive. Problems arose here in many categories:

First of all, UV map editing and material editing seem to be completely disconnected. The "visual" feedback applies ONLY to the UV editor and ignores all MATERIAL texturing, as in.. when you render you see a whole different thing. Lighting has to be activated per face, and I honestly find this ridiculous. Even when the object has been fully uvmapped using many textures in the viewport, you then need to go and create a special material with UV texture coordinate generation, for every texture used, map the texture to the right faces from the polygon editor tab (again) and assign the material, in a very tedious and complex way. To me, this disconnection between the UV editor and materials (what you see is on the screen not the material but a mix of both) is very confusing and seems to be the result that the UV editor has been created for use in the blender built-in game engine more than for 3D modelling, which is really a shame.

Maybe if custom material models were in place, or if everything was better organized, one could directly tell the material to feed its textures from the uvmapping instead of definitng a texture source (again, i mean textures, not uv coordinates, that works), thus
making everything much simpler.

Moving to animating characters, blender here has proven to be an excellent tool for the job. Animating is easy and very well done, with great results achieved in little time. Skinning was a little difficult to figure out, and the way it works is a bit strange (wish it was a bit more obvious), but i think it's not really a problem

However, when finally exporting the result to Collada, we discovered that the support for this is incomplete, and vital information such as skinning and animation was lost, or seemed to work more for importing than exporting. For some models, the exporters threw errors and died. I believe that lack of proper Collada support is vital for any small/medium game development studios, as nowadays it's becoming the export standard and replacing propertary export plugins. Even though the python api seemed complete enough for writing an exporter, It's a good deal of work because not only the exporter has to be written but also an importer. Collada handles both and is already nicely integrated to our workflow.

So, despite the problems with the UV editor, which even if i wish could be improved, the modelling tools more than compensate it's lack. I'd be urging blender developers to take Collada more seriously and please make it folow the spec as best as possible, if you want game development companies to take Blender more seriously as an alternative.

All in all, i think Blender has by a wide margin exceeded our expectations, so thanks a lot for all your work, and keep going at this great pace!

Howitzer
Posts: 0
Joined: Fri Jan 14, 2005 4:47 am

Post by Howitzer » Thu Jun 19, 2008 11:23 pm

I'm surprised no one has said anything in response yet. I think you would get much more feed back from your experience at Blenderartists.org

In any case, it's good that you are trying Blender out, and I agree with the points you've raised. Blender is good, but is not quite there yet in terms of useability. I look forward to 2.5 - in particular for the key mapping ability it is supposed to have.

I've tried many 3d software programs. Blender, Modo, and Silo are by far the best modelers out there. Maya and Max seem to neglect the modeling tools somewhat in comparison. Thats why modelers have to use plug ins to get any decent productivity out of them.

If you want to grow your studio fast, you could continue to use Blender and start hiring some of the people in this community. I'm sure many would jump at the opportunity to work.

kakapo
Posts: 0
Joined: Sat Sep 04, 2004 2:32 pm

Post by kakapo » Sun Jun 22, 2008 5:43 am

http://apricot.blender.org/

have you checked out the recent apricot builds?

if i understand your points correctly then some of those things seem to be improved now. you will be able to see the materials directly in the viewport.

i also found the material system to be one of the more obscure parts of blender but i got used to it in the meantime. :)

LetterRip
Posts: 0
Joined: Thu Mar 25, 2004 7:03 am

Post by LetterRip » Mon Jun 23, 2008 2:49 am

Hi,

thank you for your feedback.

As others have pointed out - you might find the GLSL materials preview system that was added for project Apricot to tend to meet your needs as far as texturing.

See the videos and posts at the apricot blog,

http://apricot.blender.org/

regarding Collada support - we recognize the need for more complete and robust support - the BF might do a funding drive to get a single coder supported for a 3 month period to amongst other things, improve Collada support.

LetterRip

zingbat
Posts: 0
Joined: Thu Apr 29, 2004 12:36 am

Post by zingbat » Tue Jun 24, 2008 9:14 pm

I'm very interested in what this Apricot project will become. I use Blender to make game levels but i always wished to be able to use it as a complete game creation tool. I hope we get access to the game Blender files in the end and good documentation for all the improvements.
LetterRip wrote: regarding Collada support - we recognize the need for more complete and robust support - the BF might do a funding drive to get a single coder supported for a 3 month period to amongst other things, improve Collada support.
That would be great.

Marnoch
Posts: 0
Joined: Thu Aug 28, 2008 1:14 am

Post by Marnoch » Thu Aug 28, 2008 1:24 am

LetterRip wrote: regarding Collada support - we recognize the need for more complete and robust support - the BF might do a funding drive to get a single coder supported for a 3 month period to amongst other things, improve Collada support.

LetterRip
I haven't been able to find any further reference to this and have been unable to find any recent activity on a Blender->Collada exporter anywhere else.

There are a lot of us out here who would really love to have fully functional Collada support and a lot of people would likely be willing to put up some amount of cash for it.

So, I guess what I'm asking is, is the funding drive possibility mentioned above under consideration?


Thanks,
Marnoch

joeri
Posts: 96
Joined: Fri Jan 10, 2003 6:41 pm
Contact:

Post by joeri » Tue Sep 02, 2008 4:33 pm

Solid piece of comment.

" I believe that lack of proper Collada support is vital for any small/medium game development studios "

I believe this to be true.
But If this could save you to buy 4 licenses of Max then I'd hire a python programmer to make it for you. And please share the code he/she wrote.

I'm completely lost what you mean with the material / uv bit.

Blenders UV is made for (realtime) game texturing and materials for (slowtime) renders... so yes something completely different.
Not sure what you want here... use the render materials to be baked into the realtime textures? ( I thought that was possible in blender )

Anyway... Thanks, solid piece that contains thought and time.

Post Reply