Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Solving a BIG problem (GRAPHICS) ... if it is actually a common problem

Discussion in 'General Discussion' started by GarBenjamin, Jan 4, 2017.

  1. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Hey folks... I have a question for you and I'd like to throw out an idea I have and see what you think.

    I don't know how big of a problem this really is. It could well be that I am in the minority but let me just throw it out here.

    For me, the greatest problem I face in game dev are the graphics. And I am talking about 2D art assets here. Primarily pixel art but the same applies to HD 2D art.

    Creating the art that I need to make the kind of games I want to make (and doing it in a timely manner!) has always been the greatest roadblock between a game in my head as an idea and a game completed ready for people to play.

    I've tried various things to combat this including hiring artists, trying to find existing art assets I can buy and creating all of the art myself.

    All of these cases have the problem of speed. Often I have looked for art assets wasted so much time I just drew them myself. So trying to buy existing art assets saved me no time only added wasted time.

    I've hired artists and this can be an efficient route to take once things get going. Although it can be slow up front with communication and so forth.

    And just doing it all myself is just so freaking boring & again time consuming

    So... first question is...

    Do any of you have this problem... You would be completing games regularly if you just had all of the dang art assets you need but you don't and getting those art assets is a huge obstacle?

    I've thought many times now about hiring several artists to create graphic sets... sets that will cover a variety of games with an art style and palette that will allow every piece to work together. So one could have a platformer that ends up in a shmup that ends up in a roguelike, etc.

    The problem is no art asset will be exactly what a person has in their mind. And people don't like the idea of making a game with art that looks like a dozen other games out there. I personally don't have a big problem with this because I can always edit and customize the art and add new stuff, etc.

    Anyway, so I gave this some more thought tonight and I created a humanoid pixel art character to illustrate what I am thinking.



    That's the normal size and then scaled up x2.

    So, I have put a tremendous amount of thought into this problem over the years. And what I'm thinking is how useful would it be to have art assets where everything is simple solid shaded imagery like the above? The idea being every person who bought such a pack could then recolor and add details to customize as they see fit. But all of the base images for poses and animation sequences would be provided.

    So the bulk of the work has been done. And now it is kind of like a paint-by-numbers set. And I imagine some people would just use the art "as is" which is fine too.

    See, I think this would make it so the artists could generate the art quite quickly and cheaply. And for us devs who had these packs we would have these "templates" that we could customize as we see fit for each game. Meaning we could actually take any one of these "blanks" and recolor and customize it uniquely to be used in several different games we make.

    What do you think about such an approach... would something like this be helpful in generating the 2D art you need to bring your games to life?

    For me personally, having a bunch of complete art assets done in this solid shaded style would be of tremendous value. Heck, I could prototype my games using the art "as is". And then spend time coloring and adding details at the end.

    Well thanks for taking the time to read this and I appreciate your input.

    Like I said, I don't know if maybe most everyone else just enjoys spending hours, days, weeks and months creating all the graphics from scratch. I don't. If I wasn't actually trying to make games I would enjoy it. But when trying to make a game it just becomes a huge obstacle taking so much time it removes much of the enjoyment of game dev for me.
     
    Last edited: Jan 4, 2017
    dogzerx2, zwcloud and Not_Sure like this.
  2. imaginaryhuman

    imaginaryhuman

    Joined:
    Mar 21, 2010
    Posts:
    5,834
    I guess you're thinking sort of along the lines of re-purposing the art, being able to 'get the most out of it' by making it easy to modify, making it modular etc. This modularity is a direction many people go in to try to save time. You could e.g. have custom things to hold, and mustaches ;-)

    Without you creating quite a large set of 'parts' I'm not sure how much this would appeal.

    But I do agree that in general the art content is really a difficult part of development. It's not something that can very easily be overcome with automation, or with scripting, or fancy programs that procedurally generate etc. It's one of those hard, doesn't-scale-well things that almost needs a single person to spend lots of doing working on a single image and then not being able to use it for anything else, kind of deals. I've often thought Unity would be great with some kind of super-advanced content generation system that is highly configurable/adjustable and/or can bang out variations on a theme without much tweaking needed. But this would be extremely high level and sophisticated. There are some mesh generators on the asset store that generate for example specific things like a 'plant' or a basic spaceship or something, but you can see why they are limited to specifics because it's hard to map a whole world full of possibilities into a single piece of software. Unfortunately we don't have star-trek food replicators yet.
     
    Ony and GarBenjamin like this.
  3. ShilohGames

    ShilohGames

    Joined:
    Mar 24, 2014
    Posts:
    3,015
    Even with the option to re-color the images, it would still be limited to one art style and what ever animations/poses were included. Really good pixel art games set themselves apart with unique art styles. If we listed several of the top pixel art games from last year, each of them is unique visually.

    One idea I have kicked around is finding or building some software that could convert pictures into pixel art, and have various sliders to try to mix different aspects of various art styles.
     
    theANMATOR2b and GarBenjamin like this.
  4. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Well I wasn't even thinking of mustaches, beards and such... I was coming from the angle that each developer could add whatever details they wanted. But all of the base imagery would be there, kind of like "blanks". If they wanted to give them mustaches, hair, whatever they can do it. All of the animation sequences would be there for walk, run, jump, leap, duck, knock back, using a swinging attack, using a gun, fist fighting, etc.

    The developers would add the actual weapons. I mean I guess I could easily have the artists knock out a set of weapons to be used for each of the character as well. Yeah, that might be useful would save all of us time.
     
  5. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    That is true so what I am thinking of for that is to have many (MANY) of these sets. Like a massive number of them. Say this character with all of his animation frames for everything is done and costs maybe oh I don't know like $5 or so. Then another character maybe a female same way. And various enemies same way (basically with each enemy being treated as a player character with all of the animations available).

    Of course some characters would be short and stocky, some would be super skinny, etc.

    One thing I hate about buying canned art sets is whoever made them seems to have never made a game before.

    I want to see animations for idle, walking, crawling, ducking, jumping, leaping, being injured in 2 or 3 ways (ideally a high, middle and low injury animation), climbing, swimming, picking something up, using a lever, combat in several different styles (punching, swinging, using a pistol, using a rifle, etc).

    When I say a complete set I literally mean a complete set. At least as far as what I expect from graphics assets I check out and am always disappointed by.
     
  6. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    This reminds me of back when I had an Amiga and I had a friend take dozens of photos of myself, digitized them then drew over them to make a hero character for a Castlevania style game. See that would be cool too. If someone would just put up a huge number of sets of digitized characters. Some people would use them as is I am sure but I would just draw over them. The important thing is to have all of the poses and animation sequences done. That is the tedious time consuming part for 2D IMO.
     
  7. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Just to clarify, I am not saying make this a Unity Asset with sliders that can change pants color, shirt color, etc. I mean providing all of the core of the artwork already done. All of the animations for a character done. And then the developer can take the images into the paint program of their choice and customize the images as they see fit. The benefit is they wouldn't need to worry about actually drawing all of the frames of animation and the solid shading would define each major part.

    So, it is not meant to be a game ready pack. It would be built entirely to provide a massive jump start. Because all a person would need to do is color and add details to these characters, enemies, etc. Like someone may use two shades of green and give the dude camo pants. Someone else, may use skin tones and remove the characters shirt completely. But everything would be there ready to work with. They just fill in the blanks so to speak.

    Or they give it to their artist (or hire an artist) to fill in the details. Since the actual character base designs and all images for their animations are already done the artist only needs to texture them.

    Same for lighting. Because they are provided as solid shaded people can then add lighting from any angle they like when they do their texturing.
     
  8. angrypenguin

    angrypenguin

    Joined:
    Dec 29, 2011
    Posts:
    15,617
    Are there not already rotoscoped outline sets available? I thought that kind of thing would be a common animation resource, if nothing else.
     
    GarBenjamin likes this.
  9. ShilohGames

    ShilohGames

    Joined:
    Mar 24, 2014
    Posts:
    3,015
    What is the full list for a complete set, though? For example, Stardew Valley has a lot of animations that you did not list, such as watering plants, planting seeds, jumping for joy, breaking a rock with a pick axe, etc. Every game could have a detailed and unique list of animations.
     
    theANMATOR2b and GarBenjamin like this.
  10. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    That would be worth looking for.

    I was actually just thinking... I've always thought gameplay was the most important thing but I'd still spend some minutes on drawing each image for a game. Problem is a game needs many images hundreds at the least and generally thousands when animation is taken into account.

    So... I think I have come to a decision. I am just going to knock out the graphics super fast. Like literally 30 seconds on a tree. 30 seconds on each frame of an animated character, etc.

    Then I will make the game. I should have thought of this sooner. The answer was right there all of the time. I was spending minutes per image and that was still way too long. I just need to get the time down to less than a minute per image. Ideally like 10 to 15 seconds per image. Then I can generate the content quickly enough. :)
     
  11. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Yeah you are right and see that is the issue with every asset pack that exists anywhere. There is only ever enough content to create a very limited game. I was just thinking of assets for the bulk of action, metroidvania, adventure, combat style games.

    I think there is no real solution unless several dozen people decide to do the kind of thing I was describing OR they just create packs. Hundreds or thousands of packs.

    Like I said in my post above I am just going to go a different route now. But I thought it was worth throwing the idea out for consideration.
     
  12. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    Well paradoxically, to generate lots of adaptable 2d contents ... you need 3d :D

    2D is tied to art styles, every animation need to be generated for every direction you need, you need to apply effect manually like the cutting edge technique of shadow! and many more.

    All of that come for free in 3d, animation is reusable to character with vastly different morphology, the mesh can be adapted to fit different style and rendering, you can dynamically change the appearance depending on resolution, you can modify how stuff generate relative to each others (like having eyes always on top of hair), you can mix and match part with great modularity, you have infinite resolution on animation, animation can be combine together, you change color at will ....

    and all you need is to render the result to whatever 2d you need(top down, side scrolling, isometric, you name it)!, some people even made rendering directly into good pixel art ...
     
  13. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Test complete.< 5 minutes for a character, base scene and then a tree prop.



    I can work with this. Graphics problem is solved. I still need to get faster. I need to knock this much out in about 3 minutes or less. I think that is reasonable.
     
  14. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    I was actually thinking that. I could use poser to create humanoid characters. Get 3D models for enemies and so forth and render them out. There is an asset on the store that I bought tonight to do that. But looked like it would take some time to figure it out.

    I'm just going to knock out the graphics as fast as I can while still making the objects readable.
     
  15. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    Firstly I definitely have this problem. I have a huge library of art I have commissioned, and art I have found from the numerous game art sites around the web. But yet I still have this problem. So I agree it exists, and expect its very common.

    - - -

    I think the biggest issue is that the customisation just doesn't work out that easy. As far as programmer art goes I'm not that bad, I've even managed to sell a few UI kits that I did myself. Yesterday I tried to add a gun to an animated character sheet I had. Its a fairly complex sheet about 90 frames, but low res. It took me around 2 hours to do just the idle and walk (about 10 frames), and the results were pretty average. In fact I threw them away, there's no way I was going to spend 180 (2 hours * 9 * 10 variations) for such average looking results.

    So even if you had enough animations to be useful for prototyping, and even if you had enough tiles and enemies to flesh out a reasonable game without constantly running in to things you want to do but cant due to lack of graphics, for non-artists the customisation would still either take hundreds of hours, or look bad, or both.

    - - -

    From another angle http://kenney.nl/ has 35,000 free sprites. About 12k to do with platforming. The coverage is barely enough to do any but the simplest game.

    I think this is a better approach, and if you add a well organised easy to use library with hundreds of variations and animations I think this would do a better job of solving the problem. But given you probably need something like 50,000 frames I'm just not sure its commercially viable.

    Maybe if you had a good start you could go for something like Kickstarter or Patreon.
     
    Last edited: Jan 4, 2017
    GarBenjamin likes this.
  16. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    PS 3D approach is a good one, but it does tend to limit you to a certain style.

    PPS I think you have a lower threshold of acceptable than most :)

    PPPS I just remember I tried to do something vaguely similar to this (albeit with much more styled characters):
     
    Last edited: Jan 4, 2017
    GarBenjamin likes this.
  17. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Thanks for that! I am glad I am not the only one. I see it the same way. I have looked at thousands of 2d art assets free and paid all over the Internet and even with all of those only the most basic game can be made. Exactly as you said. It is just strange to me. Either every artist making 2D packs must be thinking of ultra simplistic mobile games or is not really thinking in the true context of a game when they make these packs.

    It amazes me that even an average NES game from 25 to 30 years ago cannot be remade with the existing art assets. There would be enough to replicate about 2/3 of this level and 3/4 of that level at best. It is pretty crazy. lol
     
  18. angrypenguin

    angrypenguin

    Joined:
    Dec 29, 2011
    Posts:
    15,617
    So... would going 3D perhaps be an option? Have animated skinned meshes where you can parent accessories and re-texture components as desired, and either render them or use them to generate sprite sets. I'm sure the amortised frame time would drop significantly, then, at the cost of not having the same level of artistic freedom as you get from hand-colouring every pixel (which you don't have if you're using pre-defined shapes anyway).
     
    GarBenjamin likes this.
  19. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    That's pretty cool work! Looks like you were heading in a good direction with it.

    Ha ha about the lower threshold of acceptable... well I am at the point where I just truly don't care anymore. Well I care enough to not just scribble and call it a tree or make stick figures... hmm... stick figures.... could be a nice aesthetic... super easy to knock out. I may have to explore that in the future.

    I think I will just press on animating the character. And proceed from there. At some point I will revisit the palette and do some better colors. But for now just going to grind out art as fast as I can. I really think it is the key to making the bigger more interactive games I want to create. When animated and then recolored with color theory in mind I think it all may just look good enough to play. I probably won't have the black outlines either. Not sure yet.

    I won't know unless I try. So I am going to knock out enough graphics and animations for a decent level and then see how it turns out.

    See it is true I'd prefer to play a game with a tree looking like the one I drew above if it was interactive in some way much more than I would to play a game with a beautiful looking tree that serves no purpose other than eye candy. I am probably in the minority on that.
     
    Last edited: Jan 4, 2017
  20. RockoDyne

    RockoDyne

    Joined:
    Apr 10, 2014
    Posts:
    2,234
    ... Or you could spend a few more minutes just to ensure any potential players don't immediately gouge their eyes out. Just think that you have to prove that your game isn't made by an eight year old in the first two seconds.
     
  21. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    LOL! I hear you but problem is that kind of thinking just leads back to the same problem as before.

    That won't be the final result. Probably close but not exactly. I will touch things up at the end. Also I think of it this way I can get an answer to the question finally of just how important are graphics to gamers? Many won't play such a game I am sure. But I do wonder how many would give it a try. There is only one way to find out.
     
  22. Tusk_

    Tusk_

    Joined:
    Jan 1, 2017
    Posts:
    205
    I forsee a HUGE problem using HD vectors to create a game. For one the animations would be terrible unless you have lots of people and lots of money. Every HD Vector based art game I have seen reminds me of a malware facebook type game that I wouldn't touch with a 10 foot pole.

    Now low poly art look interesting there is a very nice art style to this, it looks appealing. The real question is how much longer does it take to animate 3D vs 2D?

    2.5D has always been great.
     
    GarBenjamin likes this.
  23. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    Why do you think so? I Never seen 3d limiting anyone to any style, I can do almost any style in 3D, given enough time for the most complex one.
    This is 3d did you notice?


    here is another



    2d is just a special case of 3d, except you don't have to draw
     
  24. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    We are specifically talking about using 3D rendered to 2D as a way of time/cost effectively creating 2D art.

    One of the most popular styles of 2D art is pixel art, making 3D render to pixel art is challenging (I've never seen it done well outside of very specifically constrained circumstances).

    Another very popular 2D style is the cel-shaded with dark outline (often seen in anime). Although there have been some good examples of this:

    a) most of these examples are still nowhere near as good as the hand drawn stuff
    b) it takes a bunch of effort tweaking models, shaders, etc to get it to work

    I'm sure anything is possible in 3D, but if it takes more time creating the art in 2D than the 3D it defeats the purpose (in this context).
     
    angrypenguin likes this.
  25. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    But you are comparing this to good artist anyway, a quick 3d renders is still above the skills of many people, it's fast, and since the render is 2d you can still go other it, either adding post process or hand correcting them. And any modification can use batch rendering so changing the size of the character mean you don't have to redo the 80+ animation in 8 directions! It's faster and better in any way I slice it. I say the reward overthrow any cost by at least one magnitude.
     
  26. Billy4184

    Billy4184

    Joined:
    Jul 7, 2014
    Posts:
    6,009
    I would never bother with sprite sheets for any 2D game i made, you simply need to make too many of them to make anything useful. And just think about it, drawing each pixel for a sprite sheet is like tapping in 1's and 0's into your computer. To animate a 3D character you just need to twist and yank some bones around.

    When you wonder why it's impossible to make the average NES game from way back, I think the issue is that the whole artistic style and workflow was not simply a reduction (technically) of what we have now, but an entirely different branch of art based on what people found to enhance the ability of a computer to display pleasant graphics. That is to say, that the tools we have now for game modelling cannot simply be turned around and used to churn out high-quality pixel graphics faster than ever before, because they're not along the same spectrum.

    On the other hand it's the perfect way to create placeholder content - the fastest way is actually to use the most modern techniques but producing simpler assets. Creating sprite sheets from animated 3D objects is a snap in Blender and could easily be produced in Unity.

    So the way I see to get to the content that you need, at least for characters, would be something like:
    • A single animation for each of the behaviours you described, running, jumping, picking up etc;
    • A highly simplified character creator with about 5% of the functionality of MakeHuman, just enough to change things like hair color, skin color, bulkiness, a single slider for skinny/built etc;
    • A sprite sheet renderer;
    • Maybe a pixelator if you want to go for that style.
    This would be extremely fast to produce a new character, you could literally move some sliders, and tap a button to bake animation sheets for each behaviour and even create a new character controller automatically.
     
  27. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    It would be of very limited use.

    What you're doing is very similar to the base sprite sheets people make for rpgmaker, for example.

    See, for example:
    Or:
    http://s244.photobucket.com/user/KIRIKA1989/media/Base Sprites/basesheet1.png.html

    However, someone who uses such sheet will be pretty much locked into your art style, if the idea is for them to draw over.

    What would be (IMO) MUCH more useful is reference pose sheet at permissive license. Basically, a huge list of poses, and frame sequences for common actions such as walking, running, etc. This could be done with stick figures, with flat shaded characters like the one you listed etc, and it could be at large resolution. Basically, something a person who's making their own sprites could open, place at the second screen and check from time to time.
     
    angrypenguin and GarBenjamin like this.
  28. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    Yes I understand the process, I've even made a little money from it: https://www.assetstore.unity3d.com/en/#!/content/11700

    But I'm going to go on believing that if we are talking about cost effectively doing 3D renders to 2D they will tend to restrict style (at least a little): I doubt I'm going to see a library of many animations in a low-res but high-quality pixel style like (for example) BroForce.

    Note that I did say I think this IS likely the right way to do it. Just that there will be some limitations.
     
  29. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    I think it is not reasonable.
    Have you seen my ludum dare entry?
    screen3.png
    See this wizard with a stick? It took about over 3 hours to make this guys with his animation set. And that's the absolute fastest speed at which I can work, while cutting every corner. I think I can shave off about 30 minutes if I automate portions of rigging process, maybe one hour, but that's it. It is very close to the ceiling in terms of how fast I can work.
    wizard.png

    Basically, I'd expect any piece of 2d art, no matter how tiny it is to take at least a hour of development time, or several. Trying to reduce it to 3 minutes (while still producing quality results with style and polish) would require artistic skill of a god. it is not reasonable by any means.
     
    wccrawford, GarBenjamin and Martin_H like this.
  30. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    It doesn't really work with it. Using 3d to make 2d art adds few issues.

    1. Wooden and "dead" movement. Basically, 3d animation (especially with "beginner" skills) produces mechanical movement that instantly stands out as something artificial. 2d art produced by a human has errors in it. Things stretch a bit, bend a bit, sizes are a bit inconsistent. 3d art has none. It produces perfect proportions. The thing is those small errors is what makes 2d art feel "alive" in the first place.
    2. Animating things in 3d takes 2x or 3x amount of time. Basically, instead of 3 parameters (x pos, y pos, rotation), you have 6 to deal with it, and 3rd dimensions and thinking about posing character in it, introduces significant slowdown to work process.

    Basically, 3d can be used to produce "referencce sheet" (like the one I mentioned), over which you can then draw over every frame. This can up the quality of your characters, and help you with complex poses, but it will not really speed up your content creation process. Also, one way to greatly increase quality of 3d characters and animations is first sketching the poses (frame by frame) with 2d gesture doodles.So your 3d editor better support doodling over the scene.

     
    theANMATOR2b likes this.
  31. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    There is also the increasingly popular solution of cutout animation, half way between 3d and 2d in term of flexibility.



    NOTHING is 2d there, nothing!

    I carefully survey the field of NPR rendering and I'm well aware of the limitation and achievement of each technique, and generally the thing is budget, it's made to cut cost so you never have a render as good (trade for another ambition), but at the same time you won't notice cg render when it's good, disney has been using them for year (well before Kuzco who use it for many scene) and nobody noticed! Outside of guilty gears, nobody has actually care pushing the technique to its limit, mostly because the thing they were going for were never to be as good as the original, but to benefit production planning (hence why some "fault" got through, they are accepted as minor cost, reigning them would negate the benefit).

    Also just because it is 3D don't mean it has to be full 3d, you still use plane attached to a 2d textures mixed with whatever 3d shape you need, making another step between 2d cutout and 3d!

    Well the idea was to help artistically deficient people to have some spritesheet, the main concern is really their own talent, not an elusive best practice that will only works for good artist, so they are limited by principle even before you enter the technique. That's not helping them, they cannot do it anyway.

    Now if you want to pick about art, nothing is impossible to dedicated artist, I have seen some very convincing example over tigsources' forum and things in between. For example that guy who obsess over ditherting made a very convincing photoshop filter to just paint it like regular color. And pixel aesthetic can be texture on low poly mesh.




    And here is an artist who turned 3d to pixel art with some degree of success, presented to the forum of pixel art high level nitpicker, http://pixelation.org/index.php?topic=18842.0:
     
    Not_Sure and theANMATOR2b like this.
  32. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    You need to realize that if someone is looking for a way to reduce asset creation time, then guilty gear xrd sign's approach definitely is not it. Their gdc slides detail their method, and there's massive amount of work involved, that is easily comparable to animating the whole thing frame by frame on paper. Basically, simplfied explanation is that their models have between 400 and 500 bones, and they are ALL animated every frame on every channel, to make it look 2-dimensional.

    The reason why the went with this approach, I beleive, was to make the game resolution independent, and reduce load on the system created by use of HD sprites (blazblue does not exactly work fast on some system).

    In those examples, every model screams that it is a 3d render. Especially the girl.

    It is not a good example for demonstrating benefits of using 3d. If anything it reminds me of Doom Troopers on sega genesis, which did not look well even at time of its release.

    Speaking of Puppet 2D and skeletally animated 2d assets, those also usually produce assets of greatly reduced quality compared to sprites which are fully hand-drawn. Basically, using this approach creates a risk of making the game look like Flash animation from 10 years ago, but in some cases there's simply no other way to go about it.
     
  33. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    It will still a step up from this:

    Which is the entire point, also there is enough free animation libraries and tools to retarget, or just modifying them, with only unity, that's a net win to someone who won't have access to a skill artist! And while the animation in 3d takes time, you get all the other thing for free, you don't have to animate shadow, you don't have to animate complex details, you don't have to animate different direction, and you don't have to animate new character (yes it's not good to have all characters with the same set of animation, but it's been done in AAA game and it's still better quality that what can be achieved by a complete noob in 2d). Those benefit add up quickly when you consider that each 2d frames need to reimplement all thoses, so in 2d a 8 directions animations will take 8 times the time to do it in 3d on stick figure (the rig) + adding all shadows and details, that's not reasonable in any way.

    EDIT MERGE POSTS

    I was answering to specific tangent question (which was quoted) about the quality of the result, it has nothing to do with the original post. ;)

    You should read better what I'm saying lol, I was saying people will attempt to perfect any technique from any other one and said it was just an attempts, and it did a lot of thing right and found some neeat tricks, it's not there yet, but other will build on it, and in fact that's what happen to guilty gears, they build on weaker attempt.

    That said in context the guilty gear was faster than their old workflow, hence whey they switch, big HD sprite was not practical anymore, and even before they were tracing 3d model to get faster!
     
    Last edited: Jan 4, 2017
  34. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    Ok thread is getting derailed, my last comment.

    a) I said "most of these examples", showing a counter example which was created by hundreds of dedicated professionals doesn't negate the statement (nor would an example created by a single individual).

    b) Your pixel examples are neither low-res nor high-quality, I specifically mentioned low-res high-quality art because this is an area where I've yet to see any reasonable solutions. People manage to make 16x16 animations fluid and expressive. BroForce art is a great example with varied characters and heaps of personality in only 32x32 pixels.

    But enough with such a pointless waste of time.
     
    neoshaman likes this.
  35. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    On a possibly more productive train, are there any freely available (ideally CC0) 3D model collections that could be the basis for rendering a very details and consistent pack?

    I tend to do a lot or searching for 2D art, not so much for 3D

    I also wonder about the licence ramifications of using say an Asset Store animation pack to animate CC0 3D models which are then used to create 2D renders that you later go on to make available for use in games. Safe side would be to get permission and specifically licence the usage.
     
    GarBenjamin likes this.
  36. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    Actually, I disagree. Try producing this image in 3d render in the same amount of time.

    I disagree on all accounts.

    Retargeting introduces a lot of issues, which you need to solve by hand, pretty much recreating animation in the process. In mecanim, retargeting can easily result in broken neck on character, for example. In Unreal 4, retargeting can result in "unsheath sword" magically turning into "cut off your own head" animation.

    You have to animate most details, unless you want your character animation look like garbage.

    And "not having to animate different direction" is compensated by having to deal with wider range of parameters, setting up ik chains. Animating Bezier-controlled IK targets is lots of "fun".

    You also lose freedom offered by 2d medium. On 2d image you can draw anything. In 3d, if you, try to animate, say, a werewolf transformation, it is a very complex project that requires planning, and making sure that your engine supports every required feature. On 2d medium, you can just draw it frame by frame and it won't take long.

    Basically... 3d does not cut as many corners as you think. For any perceived benefit, you pay a price. Sometimes the price is very high. Trying to use 3d to make 2d assets is VERY hard, and people trying to do it usually fail.

    It is not a suitable way to cut corners in asset creation. Pick 3d for visual style, not in order to try to speed things up.
     
    Martin_H likes this.
  37. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    There are a lot of sites that offer 3d content (starting with turbosquid and ending with various 3d model search engines), but generally looking for it is a big waste of time.

    Aside from dealing with licenses and making sure that you're allowed to use this or taht thing, usually you'll end up with set of models that have completely different style, and simply do not look like they belong to the same world.

    Basically, once again while looking for a way to save time, you'll spend more time looking for models than it would take for you to create your own art assets.
     
  38. JohnnyA

    JohnnyA

    Joined:
    Apr 9, 2010
    Posts:
    5,041
    If i was to do this (again) I probably would do it with custom, commissioned 3D models but use existing animations (licence permitting).

    I've had pretty good experience with retargeting, and in this context if something doesn't work you just throw it away and try a new animation. You can always buy a few custom animations to fill in any gaps.

    I think if you were aiming to get a very wide variety of content 3D models and you are happy with the "standard" 3D to 2D style this would still be the most cost effective way to get a lot of variation.

    Another choice is the Spine style animation, this will work for platformer (for example) but you would need different art for 4-way, 4 way iso, top-down, etc.

    The issue with commissioning pure 2D is a cost one, in my experience you pay per frame, and even with bulk discounts such a complete kit would be requiring thousands of frames, which would really add up. The really, really cheap "I'm just learning" kind of artists wont stick around for 10,000 frames worth of animation, so you might get 500 frames super cheap, but thats pointless if the next 500 don't match.

    (Trying to convince myself this is worth pursuing :) )
     
    GarBenjamin likes this.
  39. Tanel

    Tanel

    Joined:
    Aug 31, 2011
    Posts:
    508
    I'm not replying to your first post anymore as the thread seems to have evolved past it. I wanted to add, that maybe it could be a good idea for you to rethink pixel art and go with a style where you could leverage skeletal animation? Could potentially shave off lots of time not having to draw each frame.
     
    GarBenjamin likes this.
  40. iamthwee

    iamthwee

    Joined:
    Nov 27, 2015
    Posts:
    2,149
    Have you seen this:

     
    Ony and Peter77 like this.
  41. iamthwee

    iamthwee

    Joined:
    Nov 27, 2015
    Posts:
    2,149
  42. neginfinity

    neginfinity

    Joined:
    Jan 27, 2013
    Posts:
    13,554
    Actually, I occasionally do this while trying to draw:
    erinyes_vs_elemental.png
    tiefling crossbowman.png inkiwithgolem.png
    It helps with complex poses, lighting and perspective, but the cost of setting up this kind of prototype scene is massive, and basically by the time you're done, you start wondering whether it would've been faster to just draw the damn thing old-fashioned way.
     
    Farelle, Socrates and Martin_H like this.
  43. Martin_H

    Martin_H

    Joined:
    Jul 11, 2015
    Posts:
    4,436
    Imho it's entirely unreasonable. I think you are wasting sooo much time, trying "not to waste time on art". Maybe you should just go full-on ASCII? You obvious don't care enough about the art to aim for quality over speed, and the quickest solution to something is always "don't do it at all". Or maybe try lowres 3D, all untextured cubes?
     
    Kiwasi, theANMATOR2b and GarBenjamin like this.
  44. iamthwee

    iamthwee

    Joined:
    Nov 27, 2015
    Posts:
    2,149
    I think the conclusion is there are no shortcuts.
     
  45. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Well the thing is you're right I can't spend the time to focus on detail first & foremost. Believe me I have considered using cubes many times as well as colored rectangles.

    I consider the above graphics a step up from rectangles and cubes because I can still easily represent an object. I think anyone will be able to easily read the scene and I think it is more interesting to take on the role of a barbarian hero who looks like the above than if I had it all a solid colored rectangle (or 6 rectangle pieces for head, body, arms and legs).

    Basically I am trying to find the best way to produce graphics at high speed. And really I still think this may end up looking pretty good once the scene is alive with animation, I choose a palette for the scene to remap colors too and remove the black outline on the tree. At the time I drew the tree I wanted to see how it would look to have it outlines prominently. At the least the graphics will have consistency going for them. And I probably will add some details such as arm bands on the barbarian. The reason his arms are so thick is he is a heavily muscled dude with really big powerful arms.

    I guess I am just thinking the absolute quality of each individual piece as proportions and level of detail is not required so much as the animation, color choices and so forth are.

    I mean we've seen popular indie games literally using squares, ultra low res imagery, different geometric shapes, monochrome, silhouettes and others.

    So I think the challenge here is figuring out how to make this art work. Not to immediately rule it out. For one thing I should be able to animate the scene much more than I would if i was spending more time just creating each image. Applying color theory and remapping colors accordingly will help a lot I think.

    Truthfully I doubt I will be able to get the time down any more. I think it will take 3 minutes for each frame of humanoid animation. 1 minute for things like a tree and 1 minute for a simple backdrop like the above. And those are likely best cases but still they will be faster than I was doing spending 5 to 7 minutes on most images and up to 10 to 15 minutes on others.

    Basically it is all about striking a balance. I may need to draw hundreds or even thousands of images eventually to make the kind of games I'd like to make. Spending 15 minutes on each one would be 10 hours for 40 images. That is about a week of time. And let's face it there are always things you want to touch up.

    I don't want to go down that rabbit hole. My last little game project... Christmas Command I spent two days on it like a Ludum Dare. I also decided to put more focus on the graphics than I normally would. So I kept touching things up here and there. It ended up with the game being incomplete because spending time on graphics robbed time from the other areas of development.

    I wonder how many developers are in the same situation. How many of us have umpteen unfinished projects sitting on our hard drives because the graphics workload was just too much or was taking so much time people just burned out on the projects?

    Or how many "games" never make any headway because we all just focus on graphics obsessing over it iterating again and again. See... I was that same way many years ago. But i see it as a problem so changed my focus.

    It is probably easier for me to do this because I can appreciate a game for more than just an art showcase. And I like things that are different. Show me many games with superb HD graphics as we so often see and I'll get bored because eventually they all start looking pretty much the same. Then I see a Ludum Dare entry with very primitive art and that looks refreshing.

    I need to find a solution for this problem and this is the best solution I can see.
     
    Last edited: Jan 4, 2017
  46. iamthwee

    iamthwee

    Joined:
    Nov 27, 2015
    Posts:
    2,149
    My issue... when the original graphics are so rushed is that you'll spend tonnes of time creating different animations for these rushed graphics, the end result will be a game with 'rushed' graphics.

    I mean there's no real way to get around it.

    You could go for 2.5D I personally see this as the absolute quickest way to prototype, it will have to be a platformer though, and you could make it look decent, without creating ridiculous number of spritesheets.

    Or...Like you said you could tone the palette right down, actually, I think your first image looks pretty decent. Additionally, you can create good looking art easily, see Thomas was alone, or limbo which uses a dark colour palette, or alto's adventures.

    They're not incredibly complicated or tedious to put together, talking about the artwork and not the gameplay of course.
     
    GarBenjamin likes this.
  47. Billy4184

    Billy4184

    Joined:
    Jul 7, 2014
    Posts:
    6,009
    It sounds like you're really looking for the simplest stuff that's still pleasant, rather than really high-quality stuff - I think there should have been something made for generating this sort of thing by now. I'm pretty keen to try, at minimum I would try to make a sprite sheet renderer or something.

    In one of your last threads I mentioned trying to make something to remap colors in existing sprites. I don't think it's a very viable idea actually because in the examples I looked at artists break rules all the time for effect. So much of color and shade information is only relevant locally too - what might be a highlight in one place could be a shadow in another. Coupled with the fact that it would probably be quite hard to gather a lot of sprites that look consistent together design-wise, I decided to drop it.

    That's why I think generating everything from scratch is a lot easier - so I might try something if I have some time. I'm really keen to continue working on semi-procedural editor tools as I think that sort of thing is really necessary for us indies.
     
    GarBenjamin likes this.
  48. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Well the first image I spent 10 minutes maybe more just on that one image focusing on making a realistic (proportions, stance) male character solid shaded as a "blank". So that one character took twice as long as the scene I posted later. Texturing would add more time depending on detail maybe would add another 10 minutes. If lighting was added that'd be another 5 minutes or so.

    I've thought about Thomas was alone, Limbo and many others such as the silhouette pixel graphics games and so on. They all look good to me although I like the look of the silhouette pixel fighter game the best. It just looks more interesting to me.

    Basically I guess I want to find a solution to make Gladiator, Rastan, any arcade coin-op style game, etc with non-abstract graphics.
     
  49. GarBenjamin

    GarBenjamin

    Joined:
    Dec 26, 2013
    Posts:
    7,441
    Well thanks for all of the discussion guys. It started out on one path and ended on another but that was good. A lot of great stuff was covered.

    I'm just going to give this a go and see how it ends up.
     
    neoshaman, Martin_H and iamthwee like this.
  50. iamthwee

    iamthwee

    Joined:
    Nov 27, 2015
    Posts:
    2,149
    That's great let us know your findings!