Mari 2.0 is out
polycounter lvl 14
http://www.thefoundry.co.uk/articles/2013/03/14/491/mari-20-is-out-now/
So what's the verdict? I've never even heard of Mari until tonight when I saw it on the Gnomon Workshop Facebook page. Looks pretty neat, but doesn't 3D Coat do the same thing and for significantly cheaper? Seems to be, between that and NDO/DDO you'd have everything you need and then some.
Replies
Now that's neat - we had some Mari people giving us a demo once and PSD support was one of the issues back then. Better Layers support - we also wanted that!
Time to check out Mari again. Thanks for posting this
as for 3dcoat... well, 3dcoat does have "some" of the functionality of Mari, but Mari was designed for the movie industry before the games industry. i've found the tools it has to be very robust.
MARI is powerful but also still costly (double the price) compared to Mudbox and zBrush which also have the ability to create meshes as well as paint them. Not forgetting 3D-Coat as well, of course.
Mari has specific hardware requirements (esp. video cards). You have to have a powerful system with a lot of video and system memory if you want to have it running with other programs. I find that if I have Maya and even Photoshop running with large files that Mari tends to crash more. Also, Mari has a different file saving procedure and I still have not figured out where all the Mari files associated with a project are stored. If you need fifty 4K UV maps for a character then Mari is your only choice, but for games Mari does not really offer much more to offset the learning curve. I think the Foundry is trying to build Mari up to replace Photoshop in the movie pipeline. In games, where we typically use Photoshop plugins such as the XNormal and nDo2 and other filters--I'm not going to even consider using Mari exclusively. As noted--it is pricey
I took Justin Holt's Mari workshop. Justin released a texture painting DVD for Gnomon a few years ago (used BodyPaint3D). I asked him during a Webinar about the differences and why he now favors Mari. Most of it had to do with the movie pipeline (for example, Rattlesnake Jake in Rango, had separte UV maps for all his individual scales). That is a lot of UV maps!
I can see that if your pipeline is setup for nDo and Ddo then you may not want to throw mari into the mix, but studios working on a game where you already have a highpoly sculpt and are trying to be closer to film anyway is where I think mari really shines. Games like gears of war, the last of us, ect ect. It really helps speed the process up on any kind of hero asset. If you're doing environments and such then yeah this may not be the tool for you.
Since this is polycount, maybe it would be cool to make this discussion about what kind of features we would like to see in Mari to make it more game art friendly.
So what do you guys think? What would it take for you to use mari over photoshop for games?
No. For me personally if i had to switch to a 3d painting tool then i would need it to be actual 3d painting capable with advanced brush system like photoshop. mari has good brush system among other things but no 3d painting.
right now Mari is strictly 2D projection painting on 3d objects.
if i am going to do 2d painting, i will stick to Photoshop.
I guess I'm not entirely sure what you mean by "3d" painting. Mari gives you plenty of masking options in 3d. The projection system is very robust. I'm not sure what benefit you gain by having a 3d painting system in mari.
paint would conform to surface tangent for true 3d painting which means better brush size and brush orientation in relation to mesh. there would be no lag time between brush stroke and paint on mesh. it would reduce any texture stretching due to projection on surface that is not perpendicular to camera.
i like projection painting, it has its specific needs. however, it would be an instant sale to me if it ALSO supported real 3d painting. i just think that limited to projection painting is a NO SALE for me because as a game artist i can use Photoshop for 2d painting.
I see Mari as more appropriate for film and vfx industry since they need simultaneous painting of a huge number of textures and Mari is built for that performance.
bodypaint is projection based painting too as far as i know. i used bodypaint long long ago so anyone correct me if i am wrong.
But yeah, it would be awesome to have the option to do both.
Personally I really think Photoshop is an inappropriate tool for what we do. I'd love to replace it. What I'm torn on though is that you have stuff like 3DCoat and Mari, and then you have Substance. Texturing stuff in Substance feels like the way it should be done. But then substance has no 3D painting, or any painting actually, at all. Mari seems to nail the painting side, but then it's still layer-based, which is fine I guess. Now if Mari could read substances, then we'd really be cooking.
I prefer the 3d painting as well, but like Bigjohn said it is kinda wonky around seams.
Seams are the only reason I use projection painting really.
As for the rest, stay tuned
Sums up my feelings. At the core, what Photoshop has what all these tools don't have (yet): an extremely solid painting and layers system that's proven itself over the years. Whoever can combine this with a solid 3D paint system, a procedural system, and normal map tools wins. Right now we still have to piece together our pipeline and all tools together just don't fit the budget for one seat.
But it hinges on the paining part. I have to be able to paint 2D decals, just as well as 3D on to meshes to ditch photoshop entirely. And obviously you don't want to invest in Photoshop for cropping, resizing, and quick paintovers so any program needs basic paint functionality (or we'll use Gimp for that?
Yeah. Frustrates me more than I'd like to admit. I mean they've actually gone to the trouble of including 3D painting, and then it turns out to be utter shite.
just sayin'!
Mudbox is true 3D painting and by that I mean true painting on actual 'true' 3D geometry, not voxel based.
MARI is projection based and projection mapping and painting is technique widely used in film/tv, so it's easy to see how the tool set fits those industries.
Mudbox's painting looks simple, but it's actually very powerful (but I would say that). imo, its the closest I've seen to being like a Photoshop for 3D, some of its brushes look and behave the same, and Mudbox's layering system follows Photoshops as well. For example, blending modes and layer masks are supported and there's a Send To feature to Photoshop as well, where you can have both packages open at once.
You can even transfer paint layers between different geometry, regardless of topology, so this is very useful if you retopo a model and don't want to loose any of your work.
MARI originally came from Weta, and even though they use MARI alot, they also use Mudbox as well, for sculpting and painting.
yeah I know
Too bad it doesn't have any support for folders which makes it less then ideal for all but the simplest (or least organized) of PSDs.
>See comments in which people are hyping their own products.
Nice.
mudbox also came from weta...
I assume you mean Layer Groups?
But I don't think we're trying to recreate Photoshop, instead it's about interop. But if people think that would be a good idea for Photoshop, then suggest it: http://mudboxfeedback.autodesk.com/forums/158904-ideas-for-mudbox-forum
yes I know Mudbox started at Weta. Andrew, Dave and Tibor left to form Skymatter and take Mudbox further, and then they were acquired by Autodesk. Andrew is still with us.
Yea, I tend to think 'folders' instead of 'groups' because of the folder icon that's used, but same thing. There is actually already a suggestion for it more or less, and it's one of the top 3 most request features:
http://mudboxfeedback.autodesk.com/forums/158904-ideas-for-mudbox-forum/suggestions/2818274-improving-the-object-list
"id like to see grouping"
That's for the object list instead of layers, but I'd like to think making either of them organizable goes hand-in-hand with the other.
I'm a bit late to to the discussion. I'm the product manager for Mari. Thanks for all of the comments. This sort of discussion is really useful for us to hear.
Our aim with this release was to not only Match what could be done in PS but to enhance layered workflow wherever we could. Mask stacks, sharing and procedurals in the stack are pretty powerful.
The 'real 3D' painting question is an interesting one. For the most part people who use Mari don't tend to mention this. They are usually experienced Bodypaint and ZBrush / Mudbox artists who are used to the workflow, but there isn't much of a push to get it included. I'd be interested to hear more from people who have tried Mari and found it lacking in this respect. One nice thing is that Mari never seams.
We're very keen to hear what the games artist community has to say.
The masking system is probably the best ever in any app and I cannot thank The foundry enough to have implemented that little green arrow with the advanced masking option (Blend if function in PS). I always relied heavily on it and use it all the time in Mari. Killer product to me. Curetnly painting 8 4k 2 2k and 1 1k texture in one go...All in their own layer groups, everything with tons of layers / masks, stacks and so on. I just love it
in all honesty, i found Mari to be a very useful tool. the projection tools have always felt more robust than anything out there.
i haven't had the chance to try out Mari 2 yet, but i will be over the next couple of weeks. if the layers are as powerful as i've heard then you're certainly on to a winner.
i feel though that people are correct when they say they want a powerful 3d painting tool. it's probably the one thing that's lacking as a production tool across the board.
we have plenty of software choices that give us two of the three things we need, those being:
solid projection
solid 3d painting
solid layer system
Mari has just stepped up to the plate as two of the three, i think if any one of the software packages out there manages to get all three working together, they "win".
the reason for this opinion is that (for me at least) it's all about simplifying my workflow. removing unneeded programs where possible... why use 5 programs when i can use 3, right? as a freelancer it means i don't have to spend so much on licenses, and means i can complete tasks quicker, and more efficiently.
the day i can remove photoshop from my pipeline in favor of a truly solid 3d painting/projection tool, is a day i really look forward to. i mean... people currently even have to invest in dual monitor setups just so they can use photoshop on one screen and see their final result on another screen (on the model). having a good piece of software that does everything we need would even remove that "need" (sure people will probably still WANT two monitors, but it will remove it as a need).
i dunno i think i'm starting to ramble, i'm very tired.
i'd love to see you guys get some solid 3d painting tools on the go, but i get the impression that your primary clients are movie studios so us games devs are just the small fish heh.
I admit interaction with the 3d surface via the paint tools would be an awesome way to work. But the 2d projection painting handles a lot on it's own. Painting in context removes the step of loading things into maya or a 3rd party app like marmoset to see what it looks like as you work which is a huge leap forward.
As I delve deeper into 2.0 I'll post some more thoughts here.
Shameless plug: For anyone looking for some more info on mari used for games check the sins of a solar empire link in my sig. It's using the older version of mari so some info isn't quite up to date. For instance at the time I wasn't able to use overlapping uvs. But now you can!
Seeing it used for actual game art makes me want to give Mari another shot. I tried it before on my 1680x1050 monitor and found the UI took up almost 3/4 of the screen leaving hardly any room for the viewport. I'm running a 1920x1200 now so maybe it'll be useable
Thanks! Something I learned recently is that you don't need a lot of the UI cluttering up your workspace. J, K, L and I bring up popups at your mouse location of Color picker, Shelf, Image manager and a select-able list of /objects/shaders/channels/layers. So just feel free to stack em or close em entirely. Only keeping the ones you really use most.
Mr. Greasley this one is for you! Are there any plans for non square textures on the horizon? A lot of the projects I work on use textures of 1024 x 2048 or a similar ratio. So what I've had to do is either give mari a mesh with adjusted uvs or squash/stretch it everytime I export/import it. It's not been a huge problem but it isn't ideal.
Non-square textures are not on the roadmap for this year I'm afraid. This is however exactly the sort of feature request I'd like to hear. I'm still learning about games and games-related workflows so thanks for asking.
http://mari.ideascale.com/
... is a great place to log ideas. Although not officially run by The Foundry we watch it very closely and have closed something like 80 of the ideas off here in the last couple of years. 24 in the last release alone.
Jack
Thank you! Glad you like it. We've been working on this update for about 3.5 years. It's nice to have people start using it.
Great!
Let me know how you get on.
Try inverting the green channel?
http://vimeo.com/65408722
Is it possible to implement custom shaders already ...?
Hi,
It is currently possible (and has been since about 1.1) but we're cleaning up the API for doing so for 2.1 (in Beta at the moment)
To perform a full replacement of the shading system pre 2.1 needed a combination of API calls and editing files. In 2.1 you can replace pretty much anything you like via python.
New lighting models, adjustment layers, procedural effects and screen space filters can be written in GLSL and registered via some simple python calls.
If you're interested in this, please drop me an email at greasley@thefoundry.co.uk
Thanks
The problem with modo is that is horrible with seams and it lacks perfomance, and of course its UV system.
I don't like proyection painting, i find it pretty obsolete, it generates too many artifacts and and it's not as fast as painting in real 3D.
I have a GTX 680, is in the tested graphic cards. Gonna try it, but i don't know if the 15 days of trial will be enough, because i can not toy with games or new apps in my working days, just on the weekends. btw, 1450 for a single app is quite expensive for me. With modo there are offers sometimes.