Also at this stage let's don't judge anything. Brainstorm = no criticism. It's obvious that would be great to make at least 10% of that ideas, but let's don't kill our creativity, maybe something really interesting will born.
Greetings
What are you missing in your gamedev toolchain ?
Moderators: jesterKing, stiv
er...prob I see is quality control...it depends on the project, but usually in a game project you like to see top quality in every mesh and texture or anim...Dunno, maybe is just me. In MODs I have been , and freeware games (before I chosen this as a job, as later been hard to get home and work more on the matter.. yes, but not so focused..) , and problem nº 1 is always how to tell someone there are too many errors in a model , and is all based on no idea on proportions on the guy....Well, you don't tell him, he does for free....unless ask many times... Often you end up seing your model besides...er...other...things....Instead of only a few working on a game you could literally have a legion working on a project at the same time
Not wanting to sound like frikin' diva, but in some quality mods, or free games, they have that selection previous to picking someone. Which tends to be enough if selection method is good, or mainly, if the one who selects is already skilled artist.Anyway, imho Blender in making game art, have way essential weakness before, which are like formats for the available game engines, and lightmaps stuff, and whichever improve on character animation system... Anyway, I think is a nice idea... the "but" is my 2 cents only, however...And probably applies only to game, for something more free, non constraint, like a free comunity of mesh generation, I'd see no issues...
another idea:
it would be cool if the viewports also supported realtime shadow mapping. i am not sure but i guess once there is shader support anyway this shouldn't be too hard to do?
this could be helpful for quickly previewing shadows before baking them.
(and none game related this would enable high quality opengl preview renders of animations.)
it would be cool if the viewports also supported realtime shadow mapping. i am not sure but i guess once there is shader support anyway this shouldn't be too hard to do?
this could be helpful for quickly previewing shadows before baking them.
(and none game related this would enable high quality opengl preview renders of animations.)
My two cents on important features:
realtime IK/bones preview in the GE for demo animations for export to whatever. It’s hard to know what is needed without knowing how blender exports to crystal space.
Some general things like an easy way to demo GLSL shaders onto your objects and then export that info with your model etc.
An easy way of transferring UV’s from a high to low poly object.
Better painting tools.. Ie. proper undo’s and being able to see the size in pixels of the paint brush.. painting on multiple axis’s etc./mirror painting etc.
An editor the lets you manage multiple light baking etc. A light manager?
Easier ways of setting edges smoothness/hardness.. it must be fast and interactive. (yes yes you can use the clunky modifier thing but it's clunky.. and slow.. pls don't get offended)
The main point about any game tools developed, they must be fast and allow batch editing work flows. And when I say fast I mean Lightening fast.. the amount of time you have to spend modelling is very short. so robust and fast must be a priority.
Being able to model using QWERTY keys. faster interface. Or at least being able to set custom hot keys. (yes yes.. the blender interface is actually pretty good.. but there are too many shortcuts.. which are slow.. because the are all over the place. They need to be unified. Or at least let users set up their own shortcuts)
Mirror editing.um all I can think of right now.
oohooh.. viewing UV’s of multiple selected objects in the UV editor!
otherwise being able to edit the mesh when sculpting.
Target welding. (pls don't tell me I can press a shortcut.. I know.. using 3 clicks is slower than one. When you have a deadline.. all these uneccessary clicks add up. Again pls don't be offended there are faster ways of doing things which is what counts with game design.. trust me!)
there have been lot's of good ideas so far.. even seeing a few implented will greatly improve blender.. making it rock even more!
realtime IK/bones preview in the GE for demo animations for export to whatever. It’s hard to know what is needed without knowing how blender exports to crystal space.
Some general things like an easy way to demo GLSL shaders onto your objects and then export that info with your model etc.
An easy way of transferring UV’s from a high to low poly object.
Better painting tools.. Ie. proper undo’s and being able to see the size in pixels of the paint brush.. painting on multiple axis’s etc./mirror painting etc.
An editor the lets you manage multiple light baking etc. A light manager?
Easier ways of setting edges smoothness/hardness.. it must be fast and interactive. (yes yes you can use the clunky modifier thing but it's clunky.. and slow.. pls don't get offended)
The main point about any game tools developed, they must be fast and allow batch editing work flows. And when I say fast I mean Lightening fast.. the amount of time you have to spend modelling is very short. so robust and fast must be a priority.
Being able to model using QWERTY keys. faster interface. Or at least being able to set custom hot keys. (yes yes.. the blender interface is actually pretty good.. but there are too many shortcuts.. which are slow.. because the are all over the place. They need to be unified. Or at least let users set up their own shortcuts)
Mirror editing.um all I can think of right now.
oohooh.. viewing UV’s of multiple selected objects in the UV editor!
otherwise being able to edit the mesh when sculpting.
Target welding. (pls don't tell me I can press a shortcut.. I know.. using 3 clicks is slower than one. When you have a deadline.. all these uneccessary clicks add up. Again pls don't be offended there are faster ways of doing things which is what counts with game design.. trust me!)
there have been lot's of good ideas so far.. even seeing a few implented will greatly improve blender.. making it rock even more!
dbirb,
you are mistaken hiding works just fine inside of sculpt mode, see the wiki documentation - ctrl-shift-lmb drag rect; ctrl shift rmb drag rect; and alt h.
http://wiki.blender.org/index.php/User: ... #Modifiers
For masking someone did a patch, also another person has done a clever way to accomplish masking (but I don't recall how offhand) inside of blender without any code changes. But agreed there is no easy way to do masking at this time...
LetterRip
you are mistaken hiding works just fine inside of sculpt mode, see the wiki documentation - ctrl-shift-lmb drag rect; ctrl shift rmb drag rect; and alt h.
http://wiki.blender.org/index.php/User: ... #Modifiers
For masking someone did a patch, also another person has done a clever way to accomplish masking (but I don't recall how offhand) inside of blender without any code changes. But agreed there is no easy way to do masking at this time...
LetterRip
moody npc's
what I would like in a game is non player characters that act more natural.
I want a npc that can have diffrent moods and acts and moves acordingly.
for instance a character in normal mode walks vs a charachter in nervous mode runs . a character in tired modes sits or sleeps. The editor for this should have diffrent layers for each mood in wich there are options to change texture and and animations and maybe behavioural scripting too.
Might be much work for the artis but you can start a character with one or 2 moods and add a later time add more.
I know u can do this already by scripting alone but i think standarization would benefit so u can copy and paste behaviours from one character to the other etc.
moods that spring to mind for cool animation deviants
agressive, shy, brave, lazy, in love, drunken (good for sailors) and last but not least smart mode.
ps maybe also nicewould be to have affectors like lamps in wich you can set moods that infuence the charaters and maybe even players behaviour.
I want a npc that can have diffrent moods and acts and moves acordingly.
for instance a character in normal mode walks vs a charachter in nervous mode runs . a character in tired modes sits or sleeps. The editor for this should have diffrent layers for each mood in wich there are options to change texture and and animations and maybe behavioural scripting too.
Might be much work for the artis but you can start a character with one or 2 moods and add a later time add more.
I know u can do this already by scripting alone but i think standarization would benefit so u can copy and paste behaviours from one character to the other etc.
moods that spring to mind for cool animation deviants
agressive, shy, brave, lazy, in love, drunken (good for sailors) and last but not least smart mode.
ps maybe also nicewould be to have affectors like lamps in wich you can set moods that infuence the charaters and maybe even players behaviour.
-
DnasTheGreat
- Posts: 0
- Joined: Mon Jun 05, 2006 2:57 pm
Not sure if these have been mentioned before, but, as someone who used to use Blender to do UV, texture, and a little bit of modeling work for a game...
The Blender workflow was pretty good, from what I had to work with. My two biggest problems were tools like smoothing groups, hard/soft edges, etc. I believe Blender now has this, but it doesn't seem to be hooked into any export thing.
The second problem I had was exporters and importers. It was very very hard to get my data in and out of Blender. I believe we ended up exporting to and from obj. That had the least things lost (but still a lot was).
So, my request is to import the importers and exporters a bunch. This, to me, was by far the biggest problem using Blender in a game development environment. If possible, it'd be nice to get all those scripts to the point that
It was also difficult to get random mesh data like attachment points, etc. into Blender, but I didn't work with those much. (I mostly did the props themselves that didn't animate.)
The Blender workflow was pretty good, from what I had to work with. My two biggest problems were tools like smoothing groups, hard/soft edges, etc. I believe Blender now has this, but it doesn't seem to be hooked into any export thing.
The second problem I had was exporters and importers. It was very very hard to get my data in and out of Blender. I believe we ended up exporting to and from obj. That had the least things lost (but still a lot was).
So, my request is to import the importers and exporters a bunch. This, to me, was by far the biggest problem using Blender in a game development environment. If possible, it'd be nice to get all those scripts to the point that
- UV and mesh data is properly preserved
- The orientation is properly flipped to account for differences between Blender's axes and OTHERPROGRAM's axes
- Edge hardness modifier translated into whatever similar feature the other program has
It was also difficult to get random mesh data like attachment points, etc. into Blender, but I didn't work with those much. (I mostly did the props themselves that didn't animate.)
-
FuzzMaster
- Posts: 0
- Joined: Thu Jun 02, 2005 6:00 am
Probably asking too much, but one of the bigger hindrances when it comes to BGE is that we have to revert to old versions of concepts in order for it to work in BGE. It'd be great if things like the armature or subsurf modifier worked in game engine on the fly without having to convert it to the old, dirty way.
In the 3D paint area, I miss :
projection painting
multiple layer painting
ability to 3d paint, while keeping hidden faces... hidden ( not sure how to explain this clearly, I mean, if another object hides part of the object I'm painting on, I'd like to have the choice to see it or not. As it is now, in texture paint mode, the painted object always shows through other objects )
a crazy feature would be to be able to draw b-spline lines on the object's surface, and stroke it with the active brush...Sort of 3D vector drawing. but, ok, this is a bit crazy
my 2 cents !
projection painting
multiple layer painting
ability to 3d paint, while keeping hidden faces... hidden ( not sure how to explain this clearly, I mean, if another object hides part of the object I'm painting on, I'd like to have the choice to see it or not. As it is now, in texture paint mode, the painted object always shows through other objects )
a crazy feature would be to be able to draw b-spline lines on the object's surface, and stroke it with the active brush...Sort of 3D vector drawing. but, ok, this is a bit crazy
my 2 cents !
My 2 cents
Yes I agree with this greatly as I'm very familiar with maya and have a lot of art already created which needs to be in blender so I can use blender2crystal. The current state of collada is pretty much useless with compatibility between maya and probably 3dsmax (it chokes on materials, armatures with skinned models, armature bones with 0 length, models and armatures in the same scene but not skinned).DnasTheGreat wrote:So, my request is to import the importers and exporters a bunch. This, to me, was by far the biggest problem using Blender in a game development environment. If possible, it'd be nice to get all those scripts to the point that
The best solution has been obj for mesh/materials, collada for armatures, and bvh for animation however my maya exporter isn't the best. The very last piece of the puzzle has been a way to get vertex weight/group data from maya to blender, the closest I've got is being able to export weight data in ascii format from maya, but I have no idea on how to make a script to inject the data into vertex groups/weighting. If anyone can help me here with this, I would be GREATLY appreciated!!!
The texture baking ideas already stated are also great to have, the addition of tangent space normal map baking was really needed. For lightmaps, fortunately CS has lighter2 being worked on but a blender solution would be great for all engines. That unreal kismet thing looks a lot like what caedes was last working on for blender2crystal, using the crystal entity layer as the backend, although I'm not sure how there could be a universal solution for all engines, it seems to be a feature for a per-engine basis.
Also DnasTheGreat that edge split modifier works well now, it even picked up the hard/soft edges of my obj files automagically and exported from b2cs perfectly, so my normal-mapped models didn't have weird lighting in the sharp areas now
I didn't read every post above, but I found many interesting things.
Personally I'd like to see an improved collada support and shaders.
But I think that making blender a good tool for game development passes by a different way.
Blender is an open tool: this gives an enormous potential to it than the concurrency, when doing games.
When doing images, we're just using a tool to produce an image.
When doing games, we're using code to produce code. Having the blender code free can help *a lot* in producing a game.
I'm trying to say that blender has the potential to become not just a tool for game developing, but an entire IDE for developing games.
Developers have the possibility to make blender and their game interact in real time, this is the way.
So - to me - the best think you can do, is to make blender a tool that developers start to use with their editors at the begin of the project.
For example: since blender is already cross-platform, does a lot of interesting things and so on, a developer should start by saying:
"let's draw something (a cube "foo") into the window"
He writes down the code, compile, starts blender, open the program and voila', the cube is into a blender window.
Let's say that now we want to test a cone in the game, instead of the cube.
One shouln't go in the code and modify it.
Blender has a python interface! So the thing should be integrated, one should create the model in another blender windows, go in the python editor and write:
myprogram.getShape("foo").reset().importFromBlender("newModel");
This was just a stupid example.
Other - and more interesting examples - can be done with GL shaders, physic engines, etc etc.
Of course one developer should rely on blender functionalities while developing and unplug from it more and more going to production (i.e. writing his custom code to replace blender's).
The idea is that blender should be for developers like a framework which ease/help the writing process - and not like a modeling tool (that is anyway choosen by graphics guys).
Probably I'm just asking an integration library between blender and the game, allowing the environment control of the game (at runtime) directly in blender.
I know this is a crazy thing (talking about development time), but I really think this could be a difference, a boost, in open game development.
Personally I'd like to see an improved collada support and shaders.
But I think that making blender a good tool for game development passes by a different way.
Blender is an open tool: this gives an enormous potential to it than the concurrency, when doing games.
When doing images, we're just using a tool to produce an image.
When doing games, we're using code to produce code. Having the blender code free can help *a lot* in producing a game.
I'm trying to say that blender has the potential to become not just a tool for game developing, but an entire IDE for developing games.
Developers have the possibility to make blender and their game interact in real time, this is the way.
So - to me - the best think you can do, is to make blender a tool that developers start to use with their editors at the begin of the project.
For example: since blender is already cross-platform, does a lot of interesting things and so on, a developer should start by saying:
"let's draw something (a cube "foo") into the window"
He writes down the code, compile, starts blender, open the program and voila', the cube is into a blender window.
Let's say that now we want to test a cone in the game, instead of the cube.
One shouln't go in the code and modify it.
Blender has a python interface! So the thing should be integrated, one should create the model in another blender windows, go in the python editor and write:
myprogram.getShape("foo").reset().importFromBlender("newModel");
This was just a stupid example.
Other - and more interesting examples - can be done with GL shaders, physic engines, etc etc.
Of course one developer should rely on blender functionalities while developing and unplug from it more and more going to production (i.e. writing his custom code to replace blender's).
The idea is that blender should be for developers like a framework which ease/help the writing process - and not like a modeling tool (that is anyway choosen by graphics guys).
Probably I'm just asking an integration library between blender and the game, allowing the environment control of the game (at runtime) directly in blender.
I know this is a crazy thing (talking about development time), but I really think this could be a difference, a boost, in open game development.
#exclude <windows.h>
Re: My 2 cents
you maybe could skip the BVH step, which anyway you have problems with, and instead use *.md5. i you have the game, Doom III, installed, I think then you have a way to export models from there as md5. There's the needed stuff at doom3world.org forums, probably at editing , models or animations subforums of doom3 section.I know as I read long ago. besides der_ton made an importer that till I read, was working greatly so to import md5 animated, and weighted characters into blender, (he made also an exporter, and Max I/O as well) .So you would have in a shot both animation , mesh and weights. In my experience, the EXPORTER out of Blender and the one for Max, worked quite well (knowing a pair of things which are found in those forums. basically, in the exporter, to put th etexture name in the 'shader', as both mesh and anim files, are ascii text. )The very last piece of the puzzle has been a way to get vertex weight/group data from maya to blender,
I keep thinking game/animation formats , and support in scenery formats of lightmaps and normalmaps, is key here...
Re: What are you missing in your gamedev toolchain ?
Some more points
* headless-less mode for running scripts and do batch conversion, having the interface pop-up can get a little annoying when doing command line stuff (Xnest helps somewhat). Even better would be the abilty to 'import blender' from any Python script.
* direct support for Gimp/XCF files would make previewing texture changes easier (via xcf2png or so), in a perfect world access to layers inside the XCF would be nice, but likely not practical at the moment, maybe with XCF2
* custom object types to represent enemies, spawn-points and such when using Blender as level editor, a way to assign properties to those custom object types (health, ammo, etc.)
* some automatic way to make sure that a .blend file properly refers to external files, i.e. by relative path only, not packed. When commiting .blends to SVN its very easy to mess things up so that the .blend doesn't work properly for other users. Having a XML based representation of a .blend file should make this easy to solve this with a simple script, might also give nicer diffs in svn.
* better support for interior modeling: a way to auto-hide backfacing polygons when modeling the inside of a building or level, since they have the tendency to get in the way
* some more options in the background image dialog, so that the standard front/side/top-view images are easier setup as reference for modeling
* a way to name layers, reorder layers or in general customize layers
* a way to copy animation from one armature to another, i.e. say copy the animation of a normal size person over to dwarf (not sure if this is already possible, haven't looked very long for it)
* a way to preview an action, currently the Start/End frames of the scene are used for preview (Alt-a), but it would often be better use the start and end of an action instead when doing character animation, a function like "Play Back Animation between Markers" or so would be good, where the user could place markers that show the start and end of the action
* better ways for collaborate devepment, might not be so important in the commercial world, but in the Open Source world you might not end up having a single person doing all the animation, but multiple and having them all mess around with a single .blend file gets messy. Maybe one solution would be to not have flat .blend files, but to use directories and multiple files, i.e. each object in the Outliner could be saved to a single file. So multiple people can mess around with stuff without causing conflicts. This should allow nice tracking via SVN. To some degree this can be accomplished by linking .blend files together, but a simple "Save as Directory" checkbox in the save dialog could make things much easier.
* an "Export Library" function that works like an inverse of "Load Library" and allows to export pieces of a .blend to a new .blend file, would make some tasks easier then having to delete all unneeded stuff manually
* very minor yet annoying issue: format to save images can only be changed outside of the save dialog, not in the save dialog itself
* headless-less mode for running scripts and do batch conversion, having the interface pop-up can get a little annoying when doing command line stuff (Xnest helps somewhat). Even better would be the abilty to 'import blender' from any Python script.
* direct support for Gimp/XCF files would make previewing texture changes easier (via xcf2png or so), in a perfect world access to layers inside the XCF would be nice, but likely not practical at the moment, maybe with XCF2
* custom object types to represent enemies, spawn-points and such when using Blender as level editor, a way to assign properties to those custom object types (health, ammo, etc.)
* some automatic way to make sure that a .blend file properly refers to external files, i.e. by relative path only, not packed. When commiting .blends to SVN its very easy to mess things up so that the .blend doesn't work properly for other users. Having a XML based representation of a .blend file should make this easy to solve this with a simple script, might also give nicer diffs in svn.
* better support for interior modeling: a way to auto-hide backfacing polygons when modeling the inside of a building or level, since they have the tendency to get in the way
* some more options in the background image dialog, so that the standard front/side/top-view images are easier setup as reference for modeling
* a way to name layers, reorder layers or in general customize layers
* a way to copy animation from one armature to another, i.e. say copy the animation of a normal size person over to dwarf (not sure if this is already possible, haven't looked very long for it)
* a way to preview an action, currently the Start/End frames of the scene are used for preview (Alt-a), but it would often be better use the start and end of an action instead when doing character animation, a function like "Play Back Animation between Markers" or so would be good, where the user could place markers that show the start and end of the action
* better ways for collaborate devepment, might not be so important in the commercial world, but in the Open Source world you might not end up having a single person doing all the animation, but multiple and having them all mess around with a single .blend file gets messy. Maybe one solution would be to not have flat .blend files, but to use directories and multiple files, i.e. each object in the Outliner could be saved to a single file. So multiple people can mess around with stuff without causing conflicts. This should allow nice tracking via SVN. To some degree this can be accomplished by linking .blend files together, but a simple "Save as Directory" checkbox in the save dialog could make things much easier.
* an "Export Library" function that works like an inverse of "Load Library" and allows to export pieces of a .blend to a new .blend file, would make some tasks easier then having to delete all unneeded stuff manually
* very minor yet annoying issue: format to save images can only be changed outside of the save dialog, not in the save dialog itself
Last edited by grumbel45 on Fri Jan 04, 2008 12:46 am, edited 3 times in total.
Re: My 2 cents
Thank you for your reply oin! Der_ton seems to have a nice plugin for importing md5's, but last time I tried using md5 it didn't work too well for me. The doom3+maya "glue" function seems to only work for maya 5 and using windows doom because I tried doing it and it never worked for me with maya 8 and linux doom 3. There was a maya exporter written for maya itself and independent of doom 3 but I got some awful results when I imported the model to blender (stretched out pieces, bones were weirdly rotated which didn't help, model had seams from the UV maps so they looked a bit ugly so normal maps didn't work too well then). This could be the maya exporter or der_ton's importer, I really don't have an idea.oin wrote:you maybe could skip the BVH step, which anyway you have problems with, and instead use *.md5. i you have the game, Doom III, installed, I think then you have a way to export models from there as md5. There's the needed stuff at doom3world.org forums, probably at editing , models or animations subforums of doom3 section.I know as I read long ago. besides der_ton made an importer that till I read, was working greatly so to import md5 animated, and weighted characters into blender, (he made also an exporter, and Max I/O as well) .So you would have in a shot both animation , mesh and weights. In my experience, the EXPORTER out of Blender and the one for Max, worked quite well (knowing a pair of things which are found in those forums. basically, in the exporter, to put th etexture name in the 'shader', as both mesh and anim files, are ascii text. )The very last piece of the puzzle has been a way to get vertex weight/group data from maya to blender,
I keep thinking game/animation formats , and support in scenery formats of lightmaps and normalmaps, is key here...
*Thinks about using maya collada > 3dsmax, then 3dsmax md5 > blender* oh dear....