Amplify Texture 2 - Development Showcase

*Update: Alpha 33 now available.

​

FULL SOURCE CODE for managed code/scripts, Unity 5 support, including both metallic and specular universal shader and real-time GI support, Oculus Rift support, SLI/CrossFire support, EXR 2.0 support, Multiple camera support, solid/seamless anisotropic filtering and fixed texture addressing issues across the board specifically involving multi-VT support.

Simplified EULA, based on the standard Unity Asset Store license. The only major difference is that each Pre-Order provides you with a 3-Seat License, unlike the Unity Asset Store 1-Seat Editor Extension terms.

Amplify Texture v2.0.3 Alpha 33 - Download & Release Notes

Updated Manual - Manual Page

*Updates

Amplify Texture 2 is coming and with it a multitude of new features and improvements, we decided to create this thread in order to share all the upcoming developments.

​

As a reminder, Amplify Texture 2, unrivaled texture virtualization for Unity Pro is available exclusively in our website, currently 20% off, subject to change after final launch. Upon pre-ordering, early adopters will be granted immediate access to Amplify Texture 1.5 along with the latest development builds of Amplify Texture 2 as they become available and access to the private development forum.

Pre-Order and Product Page

Amplify Texture 2 - Development Showcase, Part 1​

In Part 1 of this series we will showcase:

• Multi-Tile UV support and automated workflow using MARI, Zbrush, Mudbox and compatible naming conventions.
• Virtualized Displacement and Tessellation.
• We take a deeper look at the already efficient variable bit rate compression used in Amplify Texture and to be improved further in the upcoming Amplify Texture 2.

Features and Workflow

Abandoned Factory Demo

Visit Amplify.pt for more information on our products and the features demonstrated.

Special thanks to The Foundry for providing the Blacksmith data set and Renato Martinho for the great accompanying music and mixing. Check out more of his work at Renato Martinho's Portfolio. The scene presentend was constructed thanks to Better Reality and their awesome scanning technology Thorskan. We used raw, high quality, high resolution scanned data in order to showcase the power of Amplify Texture 2. Thorskan is an extremely effecient system and we invite you to check it out, you might have seen it already in action in the latest reveal from Farm 51 - Get Even.

 

Looking great guys!

 

Sorry but I have to move this to the "Assets" section of the forum. Looks great otherwise

 

Ah I see, it should have been posted here. Thanks

 

No problem. I just tried clicking the links in your post but none of them would work. Website down or just glitch perhaps?

 

Interesting, we have been experiencing some slowdowns, it might be the heavy traffic load we did not anticipate. It’s working for us but thanks for the heads up we will investigate further.

 

Looks pretty cool.

 

Hi, I'm interested, but unsure if the workflow will work for me. Hopefully there will be some actual tutorials on how to take advantage of these new features before the preorder period ends.

In the videos above about uv tiling the interface you use is Mari or Modo I'm guessing? How would this work for those of us without either of these (Zbrush for example)?

 

Hello Hrothvitnir,

Thank you for your interest. We have some additional videos coming and their will certainly be some tutorials.

With Amplify Texture and Amplify Texture 2 you can use whatever workflow you require, the idea behind the tech is to be as universal as possible. For example, you can still use “traditional textures” the use of Multi-Tiled UV’s is not obligatory.

If you want to use Multi-Tiled UV’s the workflow is up to you. Amplify Texture 2 will import MARI, Zbrush and Mudbox tiles, use whatever software you have, as long as the tiles adhere to common multi-tile naming conventions it will be automatically processed.

Setting up Amplify Texture itself is extremely simple, check out the Manual for a quick overview.

Regarding tutorials on how to create Multi-Tile UV assets, there are many tutorials online that go through every step required, it’s really up to the user what method will be used.

Multi-Tile UV’s Tutorials and Info:

Digital Tutors - Maya

Maya to Mudbox - Video - Text

Zbrush MultiMap Exporter (Supports Mudbox, Zbrush, UDIM and custom naming conventions)

3ds Max Quick Overview

Modo and MARI

There is a lot of information online depending on the software you use.

Let me know if you have any additional questions.

 

Haha thanks, that's a good start for now, thanks.

 

Hiya,

Couple of questions -

1. We're doing console at the moment - would this work OK on consoles? If so, what alterations are needed so that it works with our shaders?

2. Our objective in future is a large open world. The mesh data doesn't really need to be streamed, but ideally texture data will be, but will need to be streamed in on demand - this is something unity lacks. In particular we're not needing to have textures which are huge res, but textures available when the model is visible. Can your system help in this case?

3. What performance impact or overhead is there, and is there much if any garbage collection?

 

Nice, let us know if we can provide any additional info. Although Multi-Tile UV workflows are totally abstract from Amplify Texture 2 I’m willing to give out some tips, artist to artist at ricardo@amplify.pt.

Hi,

Certainly, although some porting/adjustment work might be needed on our side depending on your console choice. We don't have a public console version but you can get more info on that by contacting us directly at info@amplify.pt. Please include the list of consoles to be used.

Regarding shader conversion, it’s a pretty straightforward process when done manually. Additionally, Amplify Texture 2 will have some virtualization presets in order to increase compatibility with 3rd parties (Skyshop, Alloy, Shaderforge etc). Conversion information is available in the Amplify Texture Manual. If you need any additional and more technical conversion info, contact us directly at info@amplify.pt and one of our engineers will get back to you asap.

Yes it can, one of the great strengths of Amplify Texture is it’s adaptability to different requirements. Even if users are not using extremely large textures it would still be beneficial to use our tech. Memory load and draw calls are reduced and overall loading times are practically non-existing.

On consoles this is an absolutely must have, we know how troublesome things can become close to a launch with all the optimizations and texture downscale. With Amplify Texture that problem is eliminated.

 

Thanks for being so open about it - we'll keep this in mind and chat further if needs be

 

Nice, looking forward to it if you decide to give it a try. Drop us an email any time

 

Our good friends at RUST LTD released Alloy 2.0, one of the best Physically Based Shading Frameworks available for Unity3d. We would like to invite all our users to try their technology.

We are very happy to announce that Amplify Texture 2 will natively support Alloy 2 and other select 3rd party frameworks. Taking advantage of cutting-edge frameworks such as Alloy will be extremely easy, in most cases as simple as selecting a Virtual Texture preset.

A new Alloy demo is also available, click the image below to forge your own blade:
​

A quick reminder, Amplify Texture 2 is currently exclusively available for Pre-Order in our website. With every Pre-Order, not only will you gain access to the upcoming private development builds you will also receive Amplify Texture 1.5.

Take advantage of the current promotional offers, Pre-Order Today!​

 

Alloy2 will supposedly have Marmoset Skyshop compatible shaders - does this mean (hopefully) that Amplify2 will support Alloy+Skyshop?

 

Absolutely, not only will it be available through Alloy, Amplify Texture 2 will also natively support Marmoset Skyshop. More news on that later on when we officially demonstrate it.

 

Hi Ricardo,

I'm extremely interested in Amplify Texture 2 for my development needs. I've tried searching through your threads here and your website as well for an answer to this question, but have yet to find it. Hopefully, you won't mind answering here:

Can you compare draw call reduction using Amplify Texture's virtualized texture to the traditional texture atlasing typically used for optimization? Size of textures and level of detail aside, is draw call reduction similar to, better than, or worse than texture atlasing? Would you recommending trying to combine both methods, or do you feel Amplify Texture's approach is sufficient?

I'm developing primarily for PC/Mac/Linux, so I'm not necessarily looking for maximum efficiency, but every little bit helps, sometimes.

Thanks in advance!

 

Amazing!
Parabéns Ricardo, Amplify é incrível!

 

Thanks for your interest in AT2. Please feel free to ask away. We'll be happy to provide any information you may require.

I might be responding to some of these technical questions, instead of Ricardo.

Definitely yes. The only problem is that we've been limited to Unity materials. So we couldn't take advantage of the implicit atlasing being done on the virtual texture. Basically, our VT atlasing is not helping draw calls right now. The plan for AT2 is to get this solved and finally starting to take advantage of the atlasing.

On the other hand, we've already implemented multi-tile UV support as another massive atlasing method to reduce draw calls. Support for static and dynamic batching is also planned.

It won't be faster in raw performance because you can't beat something as simple as basic static atlasing. It doesn't even require any extra shader calculations.

However, it ends up a better solution when combined with the streaming features; e.g. doesn't require the entire atlas texture to be in memory, virtually zero loading times, etc..

In AT1 we recommended it. For AT2, however, we recommend a Multi-tile UV approach. Hopefully, later it'll be a non-issue and the batching will just work.

AT2 we're mostly aiming at scale right now, but we're preparing to deliver massive efficiency improvements. However, if performance is critical and you're building small scale games, it might be better to rely on regular texturing. There will always be a performance hit for using Virtual Texturing, even if small.

I hope this helps. Let us know if we can assist you further.

 

Obrigado Bruno

Thanks Bruno, stay tuned @amplifycreates, much more to come.

 

Thanks for the detailed answers, Diogo! I was wondering if the VT atlasing would provide any draw call optimization, as the approach sounds similar to texture atlasing at least in theory, and I'm certainly excited to hear it is something you are planning to iron out for AT2.

What is the best way to stay informed of progress on this feature, including Multi-tile UV, as you move towards AT2's release?

Thanks again!

 

The best way is via our Blog or via Twitter but we can always keep in touch via email. We will also post all updates in this thread.

Thank you for your interest, looking forward to your creations

 

Hi Diogo and Ricardo.
I have some questions to see how much I will have to change my current texturing workflow:

-Are there any coordinate limits to the Multi-Tile UV approach?
I am planning to put all my curent UVs into the [0,0]-[16,16] uv range? that would be more than 250 uv quadrants... Would it work? What is the limit if there is any?

-Is there some sort of performance/memory penalty or wasted space on the VT when I don't use some UV quadrants while utilizing Multi-UV Tiles?
A Very badly organized example: If I layout my object to use 3 textures. The first 2 textures are neighbours and span from [0,0] to [2,1], but the 3rd texture stupidly spans from [7,7] to [8,8].
What happens exactly with the wasted space inbetween? I wonder how this works under the hood. Are this literally black spots on the VT?
Could AT2 still put other textures into this unused part of the VT, or will it just consider this space as "in use". Not that it really matters in the end when we can have 2Mx2M pixels, but I would still like to know this technical things before I start shifting all my UVs around

-And is there a new time frame for the pre order beta? Don't feel rushed ; )

Thanks in advance
Valentin

 

Hi Valentin,

There's no limit in place, except the size of the Virtual Texture. Nothing stopping you from filling an entire VT using one giant multi-tile set. Except workflow issues, for now. We'll introduce a bunch of improvements to make handling multi-tile sets much easier: support for multiple VTs per scene simultaneously visible, updating only modified tiles during editing, specific folder for tile-set caching during VT building (meant for RAMdisk), background builds (much later, but still coming), etc..

On a tile-set context, we'll skip entire missing tiles and we also skip areas of the VT which are empty. They won't take any space on disk. In the future, we'll also skip blocks that are empty; e.g. black/white diffuse areas or flat normals.

Unfortunately, however, these areas will still be marked as "used" in the VT since your control the actual VT space that gets used, even though they don't take any space on disk. So we recommend you use Rectangle or Square formats for your tile sets.

Actually, after a number of delays, we'll FINALLY be delivering devbuild1 for pre-order customers today. Unfortunately, we might end up not having a public beta. We'll skip straight to launch at the end of March, best case. Most likely April, though. On the other hand, we'll be releasing devbuilds a hell of a lot more frequently. Hopefully every week.

Promo will run until launch so there's plenty of time to get it 20% off.

Hope this helps. Feel free to query if you have additional questions.

Cheers

 

Hi Diogo.
That were the answers I was looking for.
Great to know that I can savely put my complete level into a giant, neatly-organized rectangular tile set.
I also preordered 3 weeks ago so i will give it spin in the next few weeks.

Thanks

 

Great! Thank you for you purchase. You should be receiving an email about DB1 later today then.

Cheers

 

Hi Diogo,

Will we need to convert current texture UVs on our own to leverage multi-tile UV, or will AT2 handle the conversion?

 

You'll have to supply your own multi-tile UVs. They'll have to extend over the 0..1 limit a have a range of 0,0..N,M, where N,M are the dimensions of the tile-set/grid in absolute units.

 

I currently don't have access to Maya or other similar tools for setting up multi-tile UV mapping, and even if I did, I probably would botch trying to setup the UV tile-sets.

My main concern is draw call optimization, like I posted yesterday, so as long as the intent for AT2 is to iron out the kinks with reducing draw calls via the VT, then I may not need to worry about multi-tile UVs. If, however, multi-tile UVs would be required to achieve that goal, I may need to figure out how to incorporate it into my art workflow.

Any advice or recommendations would be much appreciated.

 

Hi, I posted some info related to Multi-Tiled Uv’s on page 1, it should give you a better idea of how to set it up. Blender or any other authoring tool would be capable of offsetting Uv’s.

Well, in this case Multi-Tiled Uv’s would be required but not obligatory, we will continue to improve our methods for draw call reduction and in the future it will probably be optional depending on your type of workflow.

I’m interested in knowing more about your workflow, how do you currently create your assets and reduce draw calls? Maybe it's possible to accommodate your workflow with our current feature set.

 

I'm actually outsourcing my art assets at this time. I'm a horrible artist. Where I can, I try to leverage existing assets on the Unity store. I have also been working with a third party for contracted art as well, but for the few assets they have produced for me thus far, everything is a standalone, independent model + textures.

Ideally, I'd like to worry about the draw call reduction with respect to atlasing "after the fact", meaning I can have the assets created without worrying about the VT and multi-tile UVs ahead of time, then consolidating them within AT2. Since I am pulling assets from different sources, it's difficult to take UV tile-sets into consideration until after assets have been fully created.

With the traditional texture atlasing approach, I can worry about combining textures and UVs later on, and there are Unity tools that let you do that automatically with prefabs. You obviously don't get the benefit of the massive VT in that instance, however, so your texture size is more limited.

From what I understand, AT1's VT doesn't reduce draw calls natively with the virtualized texture, but that is something you hope to accomplish with AT2, correct?

 

I see, well you could in theory merge/atlas all the required assets into larger textures and take full advantage of Amplify Texture but I understand that it would be a bit troublesome for your current workflow.

Yes, it will certainly be explored for the best performance possible in AT2, this will be a must especially on mobile devices.

 

Great! I am eagerly watching AT2's development. There are already so many benefits to this product as-is; draw call reduction out-of-the-box would just make it even more valuable.

 

Stay tuned @amplifycreates, we will post some new info soon.

 

Hello everyone, we are very happy to announce that the second Amplify Texture 2 development build is available to all early adopters. Dev Build 2 introduces new amazing functionalities and improvements, we are getting very close to a production ready version and on track to release before June.

Amplify Texture 2 changelog:

• Multi-VT support; Up to 16 simultaneous Virtual Textures per-scene (max 64 Terapixel)
• Dramatic runtime memory footprint reduction; RAM footprint no longer depends on VT size
• Introduced Virtual Texture Hardware Level property
• Added new Cache Sizes for High and Ultra HW Level modes; 1024MB_16K and 2048MB_16K
• Improved runtime performance inside the editor

You can view the full change log in our Download Page.


We are also happy to announce one of our excellent partners; Nurulize Inc, co-founded by industry veteran; Scott Metzger, on a path to revolutionize their field of work.

Watch Scott's presentation given in the latest GTC and get a sneak peek of what is going on with the usage of Amplify Texture 2, more to come soon.

Building Photo-Real Virtual Reality from Real Reality, Byte by Byte
Click the *image to view the video.

*low resolution, captured from the presentation slideshow​

Their website will be updated soon; stay tuned Nurulize Inc.

Product Page - Amplify Texture 2

 

This is looking really great, congratulations on the hard work!

We've been looking with interest virtual texturing solutions lately and your product looks really well done and well integrated with Unity, though one area of focus for us is the overall usability of VT for mobile performance-wise.

I'm not talking of using huge photorealistic textures (such as the video above) but rather ramping up the possibilities to bypass current limitations of 8Kx8K per texture without having to load the entire texture set in memory as it would probably kill RAM requirements.

Is performance good enough on recent mobile devices (let's say 2 years or less)? Any issue with the bandwidth with tile streaming? The general idea is, can we build a game with potentially physics, navigation, animations and game logic as well as virtual texturing? Obviously it depends on a lot of factors but I'm just trying to look for potential bottlenecks due to VT running on each frame.

Last question, is mobile support available in your dev builds?

 

Thank you, much more to come, we have a lot of updates to share in the coming weeks.

It really depends on the specs of the 2 year old device but we are aiming to provide the best performance possible and removing all the current constraints related to textures. With AT2, photorealistic scenery will be possible and is one of our goals.

It’s too early to give exact estimates but we will provide all the information possible in the near future. I can't guarantee it but we are aiming at something along the lines of a iPad Mini or a Nexus 7 with OpenGL ES 3. Specs are subject to change as we are still improving the technology.

We are planning to release AT2 before June and until then performance will continue to improve.

Unfortunately the mobile target is not yet available to early adopters, but it’s coming very soon. As we mentioned above, Dev build 2 includes some major performance improvements; said improvements will greatly contribute to a better performing mobile app.

More updates soon, consider following us on twitter @amplifycreates

 

I have 600m x 600m terrain mesh. If I decide to texture it with good-looking texture density, it would take around 20GB of raw texture data. If I am correct, Amplify 2 can compress this data to something like 1-3GB? And will such terrain be viable - i.e., stream textures properly, when then player walks around normally? Like, it is really huge amount of texture data to deal. As I can see, virtual texture limit is 2Mx2M. I am not sure, if I understand those values correctly. Will that be enough to stream something like 50x50 uv tiles of 2048 size?

 

Hello Ryu,

Amplify Texture 2 can definitely help you there. Regardless of the total texture size, AT2 streams only what is required, you should have no issues with a level of that size. It would be even better if you used Multi-Tile Uv’s, no matter the number of patches in a collection it would be considered to be 1 texture, thus eliminating any unnecessary draw calls or related overhead.

In the second Amplify Texture 2 private development build we actually introduced Multi-VT support; Up to 16 simultaneous Virtual Textures per-scene (max 64 Terapixel), a major improvement from our early announcement. Texture compression is already extremely efficient and we will continue to improve it until the AT2 release.

I would recommend that you evaluate Amplify Texture 1, it should give you a pretty good idea of what you can do with our virtualization technology.

After purchasing Amplify Texture 2, users immediately receive AT 1.5, the latest development build and access to the private early adopter forum.

Amplify Texture 1.5 Trial Download

Amplify Texture 2 Product Page

 

Question, but I have played games like rage that use similiar principals from what i hear to amplify... but when i turn my head.. it kinda looks liek ass.. textures need to go from blurred to sharp.. doesnt change fast enough... and this running on a core i7 and gtx 260 at the time. I am developing a product that i intend to be for virtual reality...with oculus rift. If you turn your head quickly... it would break immersion waiting for things to load up. How well does your product perform? I have never tested it.

Can I expect issues ?

 

I've purchased your other products AmplifyColor and AmplifyMotion, both awesome Assets. I just came across Amplify Texture 2. Looks revolutionary to me.

I have a couple questions.

1. I'm creating a game where gameObjects' materials are continually being changed at runtime by player. What would be the advantage of AmplifyTexture 2 in this case? Or in other words does AmplifyTexture2 work dynamically or does it require build time pre packing setup to be useful?

2. I'm using Alloy shaders, which are super crisp. Why AmplifyTexture 2 in that case?

3. Does this work with algorithmic substances at all?

Thanks.

 

While it can happen, we do expose some tweakable params (multiple streaming threads, pages per frame, pre-pass fov) to ameliorate the issue. We've done immersive VR ourselves a while ago, using very high density scanned data. The results were very good, to be honest. In the RISE demo, the streaming was barely noticeable on the DK2.

One of the features we're trying to push for AT2 is spherical page streaming that will eliminate the issue completely. Unfortunately it is dependent of a couple of non-trivial tasks. We'll get there, however.

I recommend you give it a shot, if you have time. It should only take a few minutes to set it up and test. We'll be releasing a devbuild3 trial this week.

1) Unfortunately it relies on static packing to be feasible. You can change any params on the material except textures that are virtualized, like diffuse/normal/specular. However, you can swap materials freely.

2) We have continued support for Alloy 2 at the moment. You can build giant scenes made out of physical based materials that are compressed efficiently and get streamed automatically. No limits. Plus, AT2 will be a first class citizen in Alloy 3.

3) At the moment we only virtualize images. Baked substances will be an option. No official support yet but it's coming.

Please let me know if you require additional information.

Cheers

 

Day 2/3 #unite14 use the promo code UNITE2014 in our site and get 15% OFF all products!​

Amplify Creations Unity Pro Products

Only during Unite 2014!​

 

Last day of #unite14 use the promo code UNITE2014 and get 15% OFF all products!

Amplify Creations Unity Pro Products​

 

Announcing the first public Amplify Texture 2 development version, featuring: Third-party shader support (Alloy, Lux, Skyshop) and many other updates. We would love to hear from you, our product is close to a official release and we can really use all the feedback we can get.

The latest development version is here, Try it today!

​

 

Does amplify texture 2 have some sort of decal stamping system? All of the videos I've seen so far have been emphasizing the use of Mari and Modo, but Modo's texture painting tools are notoriously crash-happy (saying this as a longtime fan of modo, it's just a fact), and mari is prohibitively expensive.

What would interest me about AT2 is being able to texture objects normally (not using multi-tiled UVs), arrange the scene in Unity as we currently do, and add additional detail using decals that get baked down to a virtual texture (essentially how idtech 5 works). Is this how AT2 works? or do you require the user to have third party software to make your system usable?

Additionally- do you have any performance benchmarks for different machines? Rage had notorious paging issues due to virtual textures- but this was on consoles with low memory. For a PC or new-gen console would there be any noticeable texture page-in issues?

 

Hello,

Currently AT2 does not offer a stamping system; it’s definitely something we are looking into, not necessarily bound to AT2, but we are not ready to make an official announcement.

The use of Multi-Tile UV’s is not obligatory, you can still work as you always did and even combine virtualized and non-virtualized assets in the same scene.

AT2 works a bit differently when compared to ID Tech 5. The main idea is to allow anyone to take advantage of texture virtualization with the minimum amount of changes to their current workflow as possible. In short, to take advantage of AT2, you simply need to use a compatible AT2 shader, be it the Unity Standard selection or 3rd party supported frameworks such as Skyshop, Lux and Alloy. AT2 will seamlessly do it’s magic in the background, and since all texture data is virtualized you can stop worrying about technical limitations and take it a step further.

Benchmarks will be available near the actual release, we are currently working on deploying all the features we've set out to implement, and we'll start optimizing the tech during the beta stage, therefore all stats are subject to change. The hardware requirements will be quite low; our minimum supported target will be an Intel HD 3xxx, and our overall memory footprint is extremely low.

We get that question quite a few times, I believe it’s important to make a distinction between slow performance and actual errors while streaming, be it hardware or software related. Rage did seem to have it’s share of issues, especially on consoles; a fast paced FPS game is probably the worst case example for a technology like this. However, we plan to deploy spherical page detection to prevent streaming lag in this very specific case; this way, pages will only be streamed further away and only then the player moves.

What I can tell you is that if our product randomly failed to stream pages, we would not be developing version 2.

But in all seriousness, streaming textures with AT2 is very efficient and, depending on the usage case, the pros usually outweigh the costs considerably. I would recommend that you give it a try and see for yourself. We provide a fully featured unlimited-time public trial so that users can figure out if AT2 is adequate for their needs before even considering making a purchase.

 

Two questions.
Have you been able to do any testing with Unity5, is it compatible?
Do you have a ruff timeline on the spherical paging system?

Thanks Travis

 

Hi,

Absolutely, we plan to release an update during the upcoming weekend.

At the moment I don't have an exact estimate for the spherical paging system, I will have to get back to you on this one.

Talk to you soon