A good image library for PBR

Hi guys!

I started working on IBL & PBR stuff to create nice and realistic materials for my 3D models.... but I am a coder and not an artist

May be is a silly question but I need a little help: is ther any good and professional image library that can be used in Marmoset Skyshop or similar tool?

Any help is really appreciated!
Thanks
Kenshin

 

Not just Marmoset Skyshop, but also Lux, Alloy, and MOST importantly Unity5

 

Exactly!

I think/hope that "out of there" exist any good image file library that are used by professional artists...

 

Heres what I do, I got a cg textures membership, I import into the quixel suite (free right now and has like 100's of PBR textures), I generate some normals and spec maps then I do the quick PBR value and wham, PBR ready texture (sorta gotta tweak some).

 

First 2 results from Google if you're talking about IBL:
http://www.hdrlabs.com/sibl/archive.html
http://hdrmaps.com/freebies

 

They're just panoramas, he's referring to textures that are pbr ready that he can use in his game.

 

Oh, than yeah, Quixel Suite is your best bet right now.

 

Thank's for all answers.

So from what I read Quixel Suite will be absolutely under my target.

http://www.hdrlabs.com/sibl/archive.html is a good archive but I honestly do not know what image to choose to get the best effect.

Is there any good hdr/ibl map library with some selected images to produce a "good standard illumination" ready for generic metal, chrome, gold, wood, stone...?

 

That'd be a really nice place to start

 

Marmoset skyshop comes with like 8 panoramas in the examples zip folder.

 

I'm not sure I understand the question. Are you looking for cubemaps to do IBL illumination or are you looking for textures for various metals?

 

I am trying to figure out the best solution for me (coder and not artist) to realize general material ready (like wood, metal, plastic, concrete, ecc...) for PBR.

Then first of all I think is needed a good cubemap for IBL and if exist good "couple" cubemap+texture for specific material (like rusty metal) will be really great.

 

Any skybox is good for an IBL cube map, if you bake and convolute it. It only depends on the mood of the scene you want, so forget about "cubemap+texture packs", no PBR material works only with one cube map

 

Well substances will be your best bet for that, I'm sure Megascans will venture down the PBR route and I know Quixel provides for PBR. But it's still relatively new, so nothing is being set in wood or stone at the moment.

 

As mentioned a few times already Quixel will have the MegaScans Library. Which at least so far looks very awesome. GameTextures will have PBR materials soon - and it seems they are teaming up with Allegorithmic since Substance Painter is going to have a few of them in the upcoming releases.

I haven't actually used wither of them, though.

 

Try FilterForge. It makes amazing textures and has a library of over 10K that can be tweaked in a million different ways at any size you want with seamless tiling.

 

This is all way less complicated than what you may be thinking.

http://seblagarde.wordpress.com/2011/08/17/feeding-a-physical-based-lighting-mode/

I converted a regular bumped specular shader asset to pbr with a minimum of effort. Just about every conversion I've done has been super easy. Obviously the workflow is better if it's just authored the way you want it up front, but converting it is not too difficult a task.

 

If understanding how PBR works is the problem - a really easy to understand theorym can be found here:
http://www.marmoset.co/toolbag/learn/pbr-theory
http://www.marmoset.co/toolbag/learn/pbr-practice

It's a pretty comprehensive getting started guide and is written really easy to understand.

 

If you need test data only, then go the Quixel route. If you need to make your own, avoid it as you won't learn how to do it properly.

If you find yourself preferring one cubemap over another because the others "don't look right," it's because the material you've made is not physically-based. The surface should resemble what it is no matter what is being reflected.

It's worth noting that quixel megatextures are photos that have been mathematically tiled and whose reflectance values are just "theirs." Each material's roughness will have to be configured to how the engine will read it. Unreal's roughness is different from XXX engine's roughness, etc etc.. They aren't calculating the refractive index of the metals and putting that in their albedo; To put it simply: it's still a photo that's been heavily processed in the same way Crazybump processes images. Then, someone in the Quixel team is likely handtweaking all the roughness values so they are more physically accurate.

 

Well said quantum, currently there is no industry standard for pbr, especially when it comes to how the roughness/gloss/micro surface is calculated. Making it hard to make everything seamlessly work with one another

 

The Quixel web site says something about them developing a "portable HDR scanning technology" for their Megascans, but they're super vague about it. I wonder what that means.

 

While we're on the subject. Does anyone know if Unity is going to do the "traditional" PBR or if they're going to do a "metalness" style shader? I'm hoping for the former.

 

Traditional

 

We, in my other job, had private demos of the suite prior to its release and know a bit of how their scanner works. The results are incredible, but there is alot of mathematical wizardry after the photos are taken.

The biggest downside for me is that I can't modify any of the textures without the material breaking. Imaging if you wanted to move or remove a single leaf or stone pebble in the albedo. You'll have to do the same precise change to the roughness, ao, and normal maps. In this vein, megascans are very much unlike the typical cgtextures workflow of "get textures, combine them, modify them, tile them, etc." In time, just like a game using DDO *coughtitanfallcough*, you'll be able to pick the games using megascans because their use will be frequent and it has a specific look. Until Quixel decouples themselves from Photoshop, it's going to be a quirky, hacky, limited tool for AAA devs.

This is only partly why I'm going the Substance route. Substance Painter has a long way to go before it's ready for a true AAA workflow. But because it has that Substance backend it already has far more potential that DDO.

Though if Mari were cheaper, there would be no debate at all.

 

Wow, thanks for all that info. I picked up the Substance Indie bundle recently and have played with Substance Painter. I'm no artist, but I do want to play around with PBR (I have Alloy and of course PBR is soon to be a first-class citizen in Unity). Yeah, while it's neat being able to paint on a model and texture and see the PBR materials applied in real time, I'm very fuzzy on how you'd use that to do something real.

I've played a tiny bit with dDo, and at least the workflow is logical with the color-coded sections for various materials.

I watched a demo video on Mari and it looks cool. But yeah, I'm a hobbyist - $2000 is way out of my range.

 

Currently i'm just using DDO for reflectance values which I then apply to my CGTextures texture library, might use crazybump in the future to see how it compares to NDO (I really dislike NDO, find it confusing....) Crazybump also has some remove shadow filters from the diffuse which is nice, since currently I dont know the best way currently for removing shadow properly

 

I don't.. er. I understand if people have had exposure to the suite professionally but Quixel isn't 'megascans' and i don't see the megamaterial itself as needing changing and i dont even see that as remotely the point? They're elements you use with dDo for creating the full texture of a model, with pbr in mind? And i don't see the lack of industry standard as relevant, how have tools previously been more relevant?

Wouldn't it always, whatever product you use, need an element of how to set your standard basis for aspects of your model materials with the shaders in mind between everyone in the production and adjust the artwork accordingly? And wouldn't that be your experience needed to understand the shading paradigm. I don't really see Megascans as particularly important on their own as tiling materials, i see Quixel's mindset being they being used with dDo to create things that don't resemble the initial megascan material at all. I've used Mari a bunch and it's great but it's expensive as hell and just one tool of many, the same status as Quixel Suite and Substance Painter. Having them all is the right argument (imo) when it comes to authoring exactly the material you want, rather than fighting another really tedious 'x is better than y for reason reason something i dont know why im being so closed minded about this'.

The way all 3 products work is different, the results you naturally come to are different, none of them are hard to learn remotely, where is there a debate of one vs another? Especially when Quixel and Substance Painter are pocket change in the grand scheme of things

I'm happy to be corrected

In terms of just wanting a library of materials which are a tilable colour representation with a matching set of maps for pbr, Megascans looks like it could be great (It's not even out yet), i'm starting to restock my collection from Gametextures.com with the new PBR equivalents of textures, who are really great, despite their textures often being in a particular style of the artist

You can always faff about making textures from CGTextures' efforts using crazybump or knald or something but if you want access to a very large selection of tileable materials for use in texturing 3D models in which dDo and substance painter can make very quick and easy now then I like gametextures.com and the look of the up and coming Megascans

These may be unsuitable if you are all about AAA but if you are all about AAA i assume you have a team of artists and tech people writing the shaders to communicate with to make the countless materials generally needed for an AAA game to hand

 

The lack of industry standard wasn't in reference to software, but how PBR is interpreted in various engines.

Yes to both! No argument there.

They will most certainly be used for tiling materials. That's what the final output is. The mindset you mention isn't the only mindset. Their mindset is "use our tools for free, but our textures are where it's at."

You're oversimplifying. Perhaps to the lone indie it's a "whatever" choice. To studios, and I would argue for anyone, it should be a large financial one, a stability one, a workflow one, a speed one, and a features one. Your first sentence can apply to a debate between MSPaint and Photoshop.

That's great. No one is criticizing choice. All I wanted to mention is that when you go with a complete PBR material, one with multiple texture maps, and you want to modify it, it's near impossible. That's a trade-off people will have to accept.[/quote][/quote]

 

I don't think the first analogy stands at all to be honest, Mari literally does not work in the same way, or have the same deliberate direction in workflow as dDo. They're both very useful tools that can be used for different things, with dDo i'd argue being extremely suitable for very fast good looking 'rule-based' texturing and Mari being a true photoshop for 3D. For everyone and i'd think particularly the artist creating the texture a choice on how best to reach the end result they wanted is for the best, and I don't see any reason why several tools not be used for one asset. Filling out a model based on colour ids, masks and dDo's large assembly of model information (Or information you can give it like an object space normal map) quickly in dDo getting a nice effect then running it through Mari, shader channels already set up as per policy for your detail painting sounds quite lovely

Regarding not being able to change the material.. again i don't really agree, can you explain why you can't just apply the material tiled to a plane in 3Do and edit all maps of the material you want with masks using 3Do for your visualisation? Then save the changes back out as a material

And i'm definitely aware Quixel's strategy is very likely to make the big money in a LOT of Megatextures subscriptions, and I know some people will be using these tiled, but the above method could be used to alter a material on any number of tiles. It's worth noting that the software isn't free either, the beta ends in September, then you will pay. Substance Painter (you may know this) lets you paint as many maps as you want simultaneously as a core feature. Using the idea of using the right tool for the job, maybe Substance Painter could be a really intuitive place to edit a megatexture

 

Hey! I know this is an old thread, but thought I'd just mention that you could take a look at TexturePix (www.texturepix.com) for some free, super high res, high quality textures

 

Nice idea, I'll check it!