The Perils of Art Packs - LOD issues
by Arcanor · 11/13/2007 (4:02 am) · 33 comments
Unanticipated development issue: art pack standardization.
As we've been adding more and more content to Arcanoria we're running into some limitations of the model packs we've purchased. They've been written by different authors/artists, meaning there is no standardization. The model packs have always been intended to get us to the point where we can attract artists to help us with our own model designs, however at this time we have no actively dedicated modelers on our team so we're having to "work with what we've got". Some of our model packs don't have ANY LODs, which is a huge performance problem! LOD is Level Of Detail, and it's the way the engine deals with objects that are far away from the rendering camera. As an object is further and further away from the camera, the engine renders it smaller and smaller on the screen, and when it reaches a certain number of pixels the engine can render a less complex polygon mesh in its place. But those pixel levels and less complex polygon meshes must be built into the object itself. Unfortunately none of the Arteria models (which we use heavily) seem to have any LODs built in. This is a huge unanticipated performance problem, and every time our player character faces towards our skeleton dungeon (built with the Arteria dungeon construction pack) the frame rate drops immensely. On my reasonably powerful system (Radeon X1950pro, Athlon X2 4200+, 2.5 GB RAM) I'm seeing framerates drop from 30fps to 5fps just by facing in that direction. I can get over 90fps by facing the ocean. Obviously this is a priority for us to fix, but it probably means replacing/revamping all the Arteria models (and some others) with new models that have appropriate LODs, and that will mean a lot of time. Yuck!
My advice is to make absolutely sure that any model packs you buy come with a source format you can actually (and legally) edit. Many of the packs available here on the Garage Games website come with sources for 3DSMAX. But not many of us indie developers have access to a $6000 MAX license, so we're stuck using cheaper tools like Blender, Milkshape, etc. which can't read MAX format. If the model pack you're purchasing comes with alternative formats, I'd strongly suggest loading it up in your favorite 3D editor and making sure you can re-create the DTS with modified LODs and animations.
As we've been adding more and more content to Arcanoria we're running into some limitations of the model packs we've purchased. They've been written by different authors/artists, meaning there is no standardization. The model packs have always been intended to get us to the point where we can attract artists to help us with our own model designs, however at this time we have no actively dedicated modelers on our team so we're having to "work with what we've got". Some of our model packs don't have ANY LODs, which is a huge performance problem! LOD is Level Of Detail, and it's the way the engine deals with objects that are far away from the rendering camera. As an object is further and further away from the camera, the engine renders it smaller and smaller on the screen, and when it reaches a certain number of pixels the engine can render a less complex polygon mesh in its place. But those pixel levels and less complex polygon meshes must be built into the object itself. Unfortunately none of the Arteria models (which we use heavily) seem to have any LODs built in. This is a huge unanticipated performance problem, and every time our player character faces towards our skeleton dungeon (built with the Arteria dungeon construction pack) the frame rate drops immensely. On my reasonably powerful system (Radeon X1950pro, Athlon X2 4200+, 2.5 GB RAM) I'm seeing framerates drop from 30fps to 5fps just by facing in that direction. I can get over 90fps by facing the ocean. Obviously this is a priority for us to fix, but it probably means replacing/revamping all the Arteria models (and some others) with new models that have appropriate LODs, and that will mean a lot of time. Yuck!
My advice is to make absolutely sure that any model packs you buy come with a source format you can actually (and legally) edit. Many of the packs available here on the Garage Games website come with sources for 3DSMAX. But not many of us indie developers have access to a $6000 MAX license, so we're stuck using cheaper tools like Blender, Milkshape, etc. which can't read MAX format. If the model pack you're purchasing comes with alternative formats, I'd strongly suggest loading it up in your favorite 3D editor and making sure you can re-create the DTS with modified LODs and animations.
About the author
www.arcanoria.com
#22
We have and will try and take everything mentioned into account, but at least I know in our case (and I believe in others who have chimed in here) we have been more than fair in a lot of ways. I've seen single buildings on Turbosquid in a single format (Max - unusable to the average indie) without LOD, no collision mesh and outrageous sized textures going for hundreds of dollars...and that is not the exception but the rule.
Again, I know that we specifically include every art asset in formats that are Indie friendly that we produced the original models with, with our packs. If you desperately need to change something you have what you need to do it, and at a price point that the time you turn around and invest basically might be a wash, or it might make sense why it wasn't done in the first place.
Like Christophe mentioned, it's a hard road to hoe...as it sits right now, without the hours I'm going to be putting into an albeit useful and necessary update, I'm working at a dollar or two an hour. You can't do this because you plan to retire on it...you just have to love doing it.
@Stephan, part of your post was cut off here in the thread, but fortunately it came through in the email notification of this, and the answer is yes, the 40s and 60s buildings will be part of the current update I'm working on. To what extent beyond a 75% reduction in texture size, I'm not sure. Again we will have to see.
11/14/2007 (5:29 am)
@Everyone, definitely hear where you are coming from. One of the reasons Kyle and I have stuck exclusively with content for Torque is engine familiarity...and even with that we screw up things like scale and such. But we have at least ran the models in the engine...=PWe have and will try and take everything mentioned into account, but at least I know in our case (and I believe in others who have chimed in here) we have been more than fair in a lot of ways. I've seen single buildings on Turbosquid in a single format (Max - unusable to the average indie) without LOD, no collision mesh and outrageous sized textures going for hundreds of dollars...and that is not the exception but the rule.
Again, I know that we specifically include every art asset in formats that are Indie friendly that we produced the original models with, with our packs. If you desperately need to change something you have what you need to do it, and at a price point that the time you turn around and invest basically might be a wash, or it might make sense why it wasn't done in the first place.
Like Christophe mentioned, it's a hard road to hoe...as it sits right now, without the hours I'm going to be putting into an albeit useful and necessary update, I'm working at a dollar or two an hour. You can't do this because you plan to retire on it...you just have to love doing it.
@Stephan, part of your post was cut off here in the thread, but fortunately it came through in the email notification of this, and the answer is yes, the 40s and 60s buildings will be part of the current update I'm working on. To what extent beyond a 75% reduction in texture size, I'm not sure. Again we will have to see.
#23
@Alan:
We are in a looping configuration scheme:
Indie programmers can't invest too much AND indie artists can't spend the required time for "perfect" stuff without raising prices.
11/14/2007 (5:45 am)
@Tony: lots of good things on the horizon then?@Alan:
Quote:I'm working at a dollar or two an hourThat is something programmers must understand too.
We are in a looping configuration scheme:
Indie programmers can't invest too much AND indie artists can't spend the required time for "perfect" stuff without raising prices.
Quote:part of your post was cut off here in the thread8-P, thanks.
#24
11/14/2007 (7:01 am)
Which is exactly why I think most content packs are underpriced. While I appreciate a good deal like anyone, I also know when something is so low priced the artist can't possibly be making his living on it, and it saddens me because the next logical thought is that the good content packs will not continue. 
#25
One possibility for revenue for indie content creators might be "non-exclusive" customization. Let's say you have some monster models. I need a dragon a little different from what you have in your content pack - one that's more cartoony, maybe. At that point, you could offer me two options:
#1 - I could pay to have you create a new model or a mod of the existing dragon. Rate would / should be based on some kind of estimation of an hourly figure based upon the artist's speed / skill / availability and my (the customer's) needs & schedule.
#2 - I could pay a DISCOUNTED rate (half? Quarter?) for a "NON-EXCLUSIVE" customization, if the artist deems my needs are broad enough to have some marketability to the indie community. In other words, he thinks he would be able to earn back the discount (and then some) by either incorporating the mod into an existing pack (either at an increased price, or simply to make it a better value), or by productizing it all by itself.
Sure, I like the idea of a modeling tool that lets me input a few settings, move a few sliders, and press a magical "create content" button. I also want a game engine that will let me do the same for making games. But I'm not gonna hold my breath, though I expect the tools will continue to improve.
The other thing that is a challenge is that I do not even KNOW where to buy some of these content packs. I know of some of them, sure. But I was trying just this morning to remember where I'd seen that Commoners & Warriors pack (and I'd forgotten the name of it, too!). And TurboSquid - with its complete lack of standards and widely varying quality, formats, poly counts, etc. - is a terrible, TERRIBLE place to try and find stuff. But it's even harder to find indie game assets available for purchase than it is to find indie games.
11/14/2007 (8:27 am)
Content packs don't exactly have a huge market, admittedly.One possibility for revenue for indie content creators might be "non-exclusive" customization. Let's say you have some monster models. I need a dragon a little different from what you have in your content pack - one that's more cartoony, maybe. At that point, you could offer me two options:
#1 - I could pay to have you create a new model or a mod of the existing dragon. Rate would / should be based on some kind of estimation of an hourly figure based upon the artist's speed / skill / availability and my (the customer's) needs & schedule.
#2 - I could pay a DISCOUNTED rate (half? Quarter?) for a "NON-EXCLUSIVE" customization, if the artist deems my needs are broad enough to have some marketability to the indie community. In other words, he thinks he would be able to earn back the discount (and then some) by either incorporating the mod into an existing pack (either at an increased price, or simply to make it a better value), or by productizing it all by itself.
Sure, I like the idea of a modeling tool that lets me input a few settings, move a few sliders, and press a magical "create content" button. I also want a game engine that will let me do the same for making games. But I'm not gonna hold my breath, though I expect the tools will continue to improve.
The other thing that is a challenge is that I do not even KNOW where to buy some of these content packs. I know of some of them, sure. But I was trying just this morning to remember where I'd seen that Commoners & Warriors pack (and I'd forgotten the name of it, too!). And TurboSquid - with its complete lack of standards and widely varying quality, formats, poly counts, etc. - is a terrible, TERRIBLE place to try and find stuff. But it's even harder to find indie game assets available for purchase than it is to find indie games.
#26
It should provide screenshots, description, links and pricing info from the artist, as well as rating from users.
Something has been started, but it is very straight ATM.
11/14/2007 (9:53 am)
Maybe a website regrouping info about all those TGE/TGEA/TGB packs with a users review system would be good?It should provide screenshots, description, links and pricing info from the artist, as well as rating from users.
Something has been started, but it is very straight ATM.
#27
My experience has been, sadly, unless you are able to open the asset in the program in which the artist created it, you can just plan on spending some time fixing the asset up. Open a .3DS model in GameSpace, for instance? Animations don't load.. normals are flipped.. bones don't translate. I'm currently in the process of considering upgrading my GameSpace license to Truespace, and I spent several entire nights attempting to open various versions of my content pack assets insideof the Truespace evaluation, and was constantly frustrated at every turn. Mind you -- I don't blame the artists, this just the sad state of the industry.
Long and short, I'll gladly pay double for some of the content packs I purchase if I can just be assured the creators have already gone through the headache of putting the assets in game-ready format.
One other caveat about art packs I've run into is a question of style. While I'm only in proof of concept stage right now, my prototype already is a mish-mash of styles. Wherever possible, I would love to purchase assets from the same artist, in an effort to achieve a more consistent art vision. But artists tend to specialize in one area only.. which means you buy characters from one vendor, creatures from another, and buildings and flora from someone else. So I emplore artists to build "collections" as much as possible -- multiple types of assets tied together thematically. If you do, I promise you'll get my business, at least.
edit --
@Alan -- btw, I wish I had a need for a haunted house, cause the interior and exterior of that building looked incredible. I'd gladly pay twice that for an awesome looking futuristic sci-fi house. Just sayin.. ;)
11/14/2007 (4:13 pm)
I just wanted to add my "+1!" to this thread. Besides the LOD and Eula issues raised already, there's the enormous problems of compatibilty across formats, as mentioned. There are so many potentials for problems in going from one format to another, out of one tool to another, that I'm just about at the point where I won't purchase a content pack that hasn't already provided at least one set of .DTS and .DSQ exports. My experience has been, sadly, unless you are able to open the asset in the program in which the artist created it, you can just plan on spending some time fixing the asset up. Open a .3DS model in GameSpace, for instance? Animations don't load.. normals are flipped.. bones don't translate. I'm currently in the process of considering upgrading my GameSpace license to Truespace, and I spent several entire nights attempting to open various versions of my content pack assets insideof the Truespace evaluation, and was constantly frustrated at every turn. Mind you -- I don't blame the artists, this just the sad state of the industry.
Long and short, I'll gladly pay double for some of the content packs I purchase if I can just be assured the creators have already gone through the headache of putting the assets in game-ready format.
One other caveat about art packs I've run into is a question of style. While I'm only in proof of concept stage right now, my prototype already is a mish-mash of styles. Wherever possible, I would love to purchase assets from the same artist, in an effort to achieve a more consistent art vision. But artists tend to specialize in one area only.. which means you buy characters from one vendor, creatures from another, and buildings and flora from someone else. So I emplore artists to build "collections" as much as possible -- multiple types of assets tied together thematically. If you do, I promise you'll get my business, at least.
edit --
@Alan -- btw, I wish I had a need for a haunted house, cause the interior and exterior of that building looked incredible. I'd gladly pay twice that for an awesome looking futuristic sci-fi house. Just sayin.. ;)
#28
@DaveY: Dave you say that content packs are underpriced... I wholeheartedly disagree for several reasons. First, most packs are designed by people learning their trade. Almost all of the packs that I have seen have glaring deficiencies ranging from terrible to no LOD, terrible polycounts, bad scaling (IE windows and doorway sizes dont conform to the "standard" torque player size), etc.. I appreciate that you feel content packs will only get better if quality artists can make a living at it but as a direct counter point I have to add that if a person feels they should be able to make a living at something they should be very skilled, dare I even say "professional", about what they do.
Let's get down to brass tacks here. Most of the content packs that I have seen are not "retail" ready. They are meant as development tools/placeholders for advancing the detail level of your game throughout the stages of development. Of all of the people who have bought content packs how many of us have actually "released" a commercial game? IMO, if an artist wants to make a living doing game content, great. If he has the skills let him do contract work in the industry like everyone else. If he is working on his portfolio however... and expects un-funded, indie, development groups to pay their wages.. that seems kind of unfair.
This is all opinion. I dont say any of them are right or wrong and the bottom line is that the artists control what can be done with their creations and all the complaining in the world wont change it :) I think that if content creators want to get more recognition or more return financially for their stuff they need to focus a little more on making their work very "good" instead of "meh.. good enough".
In closing - There are a lot of budding content artists out here in the Torque community who do fabulous work. They are honing their skills and do not have an over-inflated sense of self regarding their work. They take feedback, they work with people to get their content even better and in the long run, IMO, those are the people that not only benefit the entire indie community but reap the rewards of such by gaining recognition for taking their skills above and beyond. This isnt to say that people doing content packs aren't skilled.. quite the contrary.. some are exceptionally skilled. But if people want to make a point about getting better content they need to hold the quality of content packs accountable to a higher standard.
One common element through this entire discussion has been the rather dissappointing level of quality in a lot of the packs that people here have purchased for one reason or another. If indie artists want the respect and financial return for their work they need to bring their "A game" and make sure their end product is worth using..
I'll leave you with this thought:
If a person spends 40 man hours working on a model and gets it to "almost retail" level it's still only worth the price of a placeholder.. If that person spends the next 10 man hours finetuning that model and polishing it so that it can be proudly displayed in any Xbox 360 game.. the value of that model just quadrupled.
EDIT:
Before anyone goes off telling me you cant purchase an Xbox 360 ready player model for a reasonable price (Like sub $200 per) do some research. I'm not talking out my hat.
http://dimension-z.net/ <<-- These guys do modelling at indie prices and their work has been used in television, advertising and many tripleA commercial games.
11/27/2007 (10:19 am)
I bought a couple of GG content packs and to be honest I feel you get what you pay for.. I think alot of people in the Torque community look at content packs as a way to make a quick buck and unfortunately the quality of the work represents that. This isn't a wild hammering rip at people who release content packs, honest :) It's just a fact.@DaveY: Dave you say that content packs are underpriced... I wholeheartedly disagree for several reasons. First, most packs are designed by people learning their trade. Almost all of the packs that I have seen have glaring deficiencies ranging from terrible to no LOD, terrible polycounts, bad scaling (IE windows and doorway sizes dont conform to the "standard" torque player size), etc.. I appreciate that you feel content packs will only get better if quality artists can make a living at it but as a direct counter point I have to add that if a person feels they should be able to make a living at something they should be very skilled, dare I even say "professional", about what they do.
Let's get down to brass tacks here. Most of the content packs that I have seen are not "retail" ready. They are meant as development tools/placeholders for advancing the detail level of your game throughout the stages of development. Of all of the people who have bought content packs how many of us have actually "released" a commercial game? IMO, if an artist wants to make a living doing game content, great. If he has the skills let him do contract work in the industry like everyone else. If he is working on his portfolio however... and expects un-funded, indie, development groups to pay their wages.. that seems kind of unfair.
This is all opinion. I dont say any of them are right or wrong and the bottom line is that the artists control what can be done with their creations and all the complaining in the world wont change it :) I think that if content creators want to get more recognition or more return financially for their stuff they need to focus a little more on making their work very "good" instead of "meh.. good enough".
In closing - There are a lot of budding content artists out here in the Torque community who do fabulous work. They are honing their skills and do not have an over-inflated sense of self regarding their work. They take feedback, they work with people to get their content even better and in the long run, IMO, those are the people that not only benefit the entire indie community but reap the rewards of such by gaining recognition for taking their skills above and beyond. This isnt to say that people doing content packs aren't skilled.. quite the contrary.. some are exceptionally skilled. But if people want to make a point about getting better content they need to hold the quality of content packs accountable to a higher standard.
One common element through this entire discussion has been the rather dissappointing level of quality in a lot of the packs that people here have purchased for one reason or another. If indie artists want the respect and financial return for their work they need to bring their "A game" and make sure their end product is worth using..
I'll leave you with this thought:
If a person spends 40 man hours working on a model and gets it to "almost retail" level it's still only worth the price of a placeholder.. If that person spends the next 10 man hours finetuning that model and polishing it so that it can be proudly displayed in any Xbox 360 game.. the value of that model just quadrupled.
EDIT:
Before anyone goes off telling me you cant purchase an Xbox 360 ready player model for a reasonable price (Like sub $200 per) do some research. I'm not talking out my hat.
http://dimension-z.net/ <<-- These guys do modelling at indie prices and their work has been used in television, advertising and many tripleA commercial games.
#29
Also, how can you possibly tell if you want to use these packs without opening them up? It is a huge waste of time and money to find out something has no LOD models and has a ridiculous polycount, or that you can't alter a texture that completely clashes with your hue and saturation levels in your game.
12/22/2007 (6:43 pm)
Thanks for the critique. To me, these clauses in the EULA make them completely useless. The only one that is reasonable to my thinking is the one game limitation. Especially if the EULA is not made clear before you buy the pack. My partner picked up a couple of the most popular packs, and tells me it's all completely useless to him anyhow. Not everyone is lucky to have a great modeler on staff, but the idea of opening up a content pack and then getting a ridiculous EULA like that annoys me a great deal. Also, how can you possibly tell if you want to use these packs without opening them up? It is a huge waste of time and money to find out something has no LOD models and has a ridiculous polycount, or that you can't alter a texture that completely clashes with your hue and saturation levels in your game.
#30
How hard is it to make the LOD yourself? Do you just save the file you have in a lower quality level in any art program there is? Does it require a lot of work? I would think most things can be saved in various quality levels quite easily by the artist, when he goes to save it in that format. I'm going to post over at the Arteria forum and ask.
Its only fair that if you have a problem, you let the person you have a problem with know. Since he does not develop in Torque, perhaps he was unaware of the problem.
12/28/2007 (3:53 pm)
Did you post your complaint about the Arteria packs on their official forum? I notice when people complain about things, sometimes they get changed.How hard is it to make the LOD yourself? Do you just save the file you have in a lower quality level in any art program there is? Does it require a lot of work? I would think most things can be saved in various quality levels quite easily by the artist, when he goes to save it in that format. I'm going to post over at the Arteria forum and ask.
Its only fair that if you have a problem, you let the person you have a problem with know. Since he does not develop in Torque, perhaps he was unaware of the problem.
#31
In any case, we started down the road of making the LODs ourselves, but it was an unanticipated need, and our art resources are very limited. My art guy then decided it would take less time to produce all new models to replace the Arteria dungeon construction kit stuff. So that's what we did, and the new models look great, AND have appropriate LODs.
My point in bringing it up here was only to say that when dealing with purchased art packs (not just Arteria), we have had to spend a lot of unanticipated time doing things that I took for granted but found were missing, or done incorrectly, or in a non-standard or non-optimal way.
12/28/2007 (6:29 pm)
@Wolf Dreamer: Why would you assume I'm not being "fair" about it? Trust me, he's plenty aware of the issues (I correspond directly via emails), and he's not given me satisfactory answers as to how he plans to resolve them. I WAS, however, trying to be somewhat discreet in this blog. But I see you've gone and posted there on the Arteria forums already. So much for trying to keep this low key. Gee, thanks.In any case, we started down the road of making the LODs ourselves, but it was an unanticipated need, and our art resources are very limited. My art guy then decided it would take less time to produce all new models to replace the Arteria dungeon construction kit stuff. So that's what we did, and the new models look great, AND have appropriate LODs.
My point in bringing it up here was only to say that when dealing with purchased art packs (not just Arteria), we have had to spend a lot of unanticipated time doing things that I took for granted but found were missing, or done incorrectly, or in a non-standard or non-optimal way.
#32
Regarding LOD of textures, i know for sure that this really helps performance in game, as i do the same with dts shapes. I ran some tests whereby i had 3 lod variations on foilage i had made in dts, whereby i reduced the texture size at 512, 256, 128, sometimes even going down to 64, and FPS soared, whilst removing the lod of textures affected performance hugely.
The reason i have never implemented lods on my saleable objects, is my one complaint with the way torque handles LOD, in terms of pixel size and not camera distance, therfore LOD can vary alot if people are playing on a low resolution, to those playing the same game with a higher resolution setting, therefore how does one determine the correct LOD for an individuals game requirements.
If it was don on camera distance away from an object, then it wouldnt matter on which screen resolution the game was played in, and thus would make the LOD process easier to determine.
Any thoughts would be helpful on this
Steve
01/08/2008 (3:16 am)
Hi Regarding LOD of textures, i know for sure that this really helps performance in game, as i do the same with dts shapes. I ran some tests whereby i had 3 lod variations on foilage i had made in dts, whereby i reduced the texture size at 512, 256, 128, sometimes even going down to 64, and FPS soared, whilst removing the lod of textures affected performance hugely.
The reason i have never implemented lods on my saleable objects, is my one complaint with the way torque handles LOD, in terms of pixel size and not camera distance, therfore LOD can vary alot if people are playing on a low resolution, to those playing the same game with a higher resolution setting, therefore how does one determine the correct LOD for an individuals game requirements.
If it was don on camera distance away from an object, then it wouldnt matter on which screen resolution the game was played in, and thus would make the LOD process easier to determine.
Any thoughts would be helpful on this
Steve
#33
But it makes sense that if you are double the screen resolution, you can double the distance before an LOD switch is noticeable.
01/08/2008 (10:34 am)
Actually, I really prefer the way Torque handles things. Pixel resolution is a more accurate measure of LOD visual tolerance than camera distance. I'm constantly telling my artists that a particular model DOES NOT NEED 1000 polygons when it is only 16 pixels high. But it makes sense that if you are double the screen resolution, you can double the distance before an LOD switch is noticeable.
Torque 3D Owner Tony Richards
Like I said, I don't blame you or any of the other artists working hard to create beautiful content packs. It's the tools that don't meet the needs, not the artwork.
What if 3ds Max, Maya, Blender, Milkshape and other tools had a plugin that supported the file format I mentioned previously? What if tools existed that made it possible for you to easily create morph targets and multiple levels of detail?
That day is coming... I can't speak for the other tools, but give us another year and IndieZen will close to achieving this goal. And it'll be well under $1500.