Messages

106

[Max] General Discussion / Re: AI super-resolution speedup

« on: 2019-04-14, 02:17:36 »

AI upscaling is much better than pretty much any other upscaling algorithm out there. Opening the possibility of a feature where you would actually render at a much lower resolution and 4x faster (or even more), while looking pretty much the same as a higher resolution render.

Again, I'm not saying this feature would be useful for final renders. As even the animation scenario described above would be covered by current plugins, as you only need to set up and run once.

Maybe you work in a different way, but when I'm working on a new scene, I don't need to see 100% of the details all the time. If I can get a 95% approximation that allows me to place and adjust lights, adjust materials etc... it can speed up my workflow significantly, the same way Nvidia Denoiser did.

107

[Max] General Discussion / Re: AI super-resolution speedup

« on: 2019-04-13, 15:59:19 »

I will post some images later.

I didn't imagine to be meant for final images, but for drafting... however, it would be great for final animations. In my tests, upscaling 1920 to 4k is kind of the best scenario, it's hard to see the difference. 

Upscaling 960 to 1920 does produce some artifacts, but I still think it would be useful in the same way Nvidia Denoiser is. 

108

[Max] General Discussion / AI super-resolution speedup

« on: 2019-04-12, 23:18:38 »

Nvidia denoiser has made a huge impact in our workflow, allowing us to vizualize a close aproximation of the final render much faster.

With the latest set of supersampling algorithms, I think there is another oportunity to bring even more speed into the process.

My idea is to have a checkbox called SuperSampling that would essentially devide the render size by half and use AI to upscale to the expected resolution.

I'm not sure, but I believe supersampling is much faster than actually rendering 4x the pixels.

This won't be intended for final renders. But it could be awesome to speed up work.

I've done a few tests with https://topazlabs.com/gigapixel-ai/ , but having it integrated in corona would be amazing. What do you think ?

 

109

Hardware / Re: Threadripper 2990WX

« on: 2019-03-20, 21:21:35 »

I would recomend you do more research on RAM.

Threadripper (and it looks like specially corona) are extremely sensitive to ram frequency / # of channels.

I'm too lazy to do a full research but I can give you a summary of my experience using 128gb of Corsair Vengance memory with the 2990wx.

Although I have 3200MHz memory, I couldn't get it to work over 3000 MHz (again, to lazy to tweak and tweak)

These are Corona Benchmark results as a refference.

Single stick at 3000MHz ( 1:12)
2 sticks at 3000MHz (1:08)
4 sticks at 3000MHz (0:38)
8 sticks at 3000MHz  (Blue screen :( )
8 sticks at 2166MHZ (1.06)

As you can see, you need to populate 4 channels to reach the full potential of the processor (not a problem since you are planning to populate all of them... however, frequency also has a big impact)

I'm not sure why I couldn't make all 8 sticks to work at 3000, someone said the corsair vengance might not me the most compatible modules.

My recomendation would be to get 3200MHz momory and reasearch which brand works best to ensure you can run 8 of them at that freq.

110

[Max] Feature Requests / Re: The most wanted feature?

« on: 2019-01-01, 22:58:58 »

I've been on both sides of product development and let me assure you a few things.

-Decisions are unimaginably more complex for the corona team that what you think
-They spend 100x more time thinking about these issues than any customer ever will.
-And based on what I've seen, the Corona team is also working very very hard and they have absolutely no reason not to build the best product they can.

I know, as a customer, there is a small set of features that are disproportionately more important because of your specific use case, and it can be very frustrating when the product priorities don't align with yours.

Just remember that there are hundreds of use cases and hundreds of "hidden" complexities in running a business and building a product.

This is not to say that we as customers should shut our mouths and just wait for the dev team to do their best, but I find that complaining is not the most useful thing we can do, here is what I think it's better:


- Understand WHY a feature is not here, sometimes it's not used by enough people, sometimes it's too complex to build, and in some rare cases, the feature is being overlooked by the dev team.
- Build a strong case on why a certain feature is important and try to get more people to support the idea and help the dev team see something they might be missing (like a market trend, new tools, new workflows etc.)

In the end, I don't think neither this poll nor the trello should be used to inform future jobs we take, we should use them as tools that reflect vision and trajectory, as I currently interpret both of these are:

The poll: People are liking the results from Fstorm and other GPU renders and would want to have those goodies in Corona too. (geo pattern, tone mapping, gpu) Second, the industry seems to be moving to PBR and people are starting to miss proper PBR support in Corona.  Most of the other items seem to have niche support and not general support.

The Trello: Seems like the Corona team is focusing on closing on "essential" features before jumping to new features: stable light solver, caustics, clearcoat, volumetrics. These are MUST have features for any production-grade render. This is a sign of maturity for the product. 
 
I don't expect they will focus in too many new features until Corona is stable and robust as a production render, something a studio can pick up knowing that it will work for them 99% of the time and they won't need to hack or fake stuff out, or switch renders in a few years. Once that is solid as a rock, I would expect they would start exploring things.

Exploring things means breaking things, and when client's livelihood depends on your product, you can't be as aggressive with new features. Fstorm can get away with whatever they do because pretty much no one is using it for "serious" work.

111

General CG Discussion / Re: Photogrametry with few images

« on: 2018-12-06, 23:06:24 »

Not a bad idea, the problem is there are 200 different subjects. Any idea on how to approach that?

112

General CG Discussion / Re: Photogrametry with few images

« on: 2018-12-06, 20:31:53 »

I'm actually more interested in mapping that model quality or detail.

I have some experience with photogrammetry, we use Capture Reality. I know all the reasons why this won't work, but still, I've gotta keep trying :)

Thanks for the tip on the Facebook group, I will post there and see if I get lucky.

113

General CG Discussion / Photogrametry with few images

« on: 2018-12-06, 19:42:26 »

Hi,

I have an odd problem and maybe some expert here can help me get un-stuck.

I need to do the best possible 3d reconstruction of a human body based only in 5 pictures taken at 90, 45, 0 -45 and -90 degrees.  Before you ask, no, I can't take more pictures, unfortunately, as I'm using an existing dataset :(

Of course, I know the reconstruction won't be great or maybe even good, based on the extremely limited data set.

Some requirements:

-There are 200 human sets, so doing a manual 3d model for each one is not feasible, but doing some manual work is reasonable (like manual alignment, manual point matching etc)
- The 3d model should be textured too.
- The pictures have plenty of details and texture
- The pictures where captured with a filter that removes all reflections so the texture detail is pretty solid and there are no specular occlusion
- the 90 and 45 degree images are not precisely at 90 and 45, but close
- I only need to build the front of the body
- The camera FOV is known

I've done research using many photogrammetry software and could not find one that has a special "low number of images" mode. Maybe a human-assisted mode?

Hope someone here can help.

Thanks :)

114

[Max] General Discussion / Re: Overblown light

« on: 2018-10-30, 19:49:24 »

I tried with 255 7 1, I get the yellow highlights on objects and white blown light. Closer, but fstorm still looks better.

Can anyone tell me if this is a "problem" on Coronas side, or if this is by design (e.g. this output is better for compositing) ?

And if it is a problem, if there is a solution in the works and what is the solution?

In my day to day, I've never encountered this scenario, and I don't really need this to be fixed. Just curious that maybe this has to do with why Fstorm renders usually look more photographic ?

Oh and BTW, thanks for your answers  guys :) I really appreciate you taking the time to provide clear and concise answers instead of a lecture on why I SHOULD be an expert in X or Y to use corona.

115

[Max] General Discussion / Re: Overblown light

« on: 2018-10-30, 18:49:31 »

I used Kim Amlan... graphic2 lut, which is my favorite. Which one do you recommend ?

116

[Max] General Discussion / Re: Overblown light

« on: 2018-10-30, 17:23:13 »

Does that "for now" means a change is planned?

117

[Max] General Discussion / Overblown light

« on: 2018-10-30, 05:28:25 »

Hello,

I know the answer to this question is already in the forum, probably in dubcat's hideout or something... but the subject is far too technical for me.

So I was watching this video

and wanted to try in corona. I immediately recognized the issue as something I've glanced on the forums but have never been willing to fully understand.

But I imagine, there must be others like me, that just would like to know a simple way (if it exists) to avoid or overcome this issue.

I tested with Fstorm and it seems to handle it better.

I know this is related to Linear workflow, ACES, sRGB etc... I'm not looking for the complicated answer. I just want the best possible render with the least possible tweaking, preferably straight out of the render.

Thanks :)

 

118

General CG Discussion / Re: Toon / painterly shader

« on: 2018-10-02, 16:05:48 »

I would use https://topazlabs.com/impression/ on top of the final render.

Toon is such a niche use case that I've never seen any render engine do an elegant job at it.

I'd rather have the corona team focus on bigger features and use a specialized software to get the toon look

119

[Max] I need help! / Re: Massive 90k pixel rendering size

« on: 2018-08-16, 21:00:21 »

This might help you.

https://topazlabs.com/ai-gigapixel/?utm_source=Topaz+Updates&utm_campaign=a39ca50223-AI_Gigapixel_CAMPAIGN_2018_08_16&utm_medium=email&utm_term=0_5e4dd8652a-a39ca50223-90364373&mc_cid=a39ca50223&mc_eid=a237073fa9

120

[Max] I need help! / Re: Massive 90k pixel rendering size

« on: 2018-08-16, 03:45:45 »

The way I handle ignorant clients requesting resolutions that make no sense...

Render at half or even 1/3, then scale in photoshop.

I find that a huge number of designers don't understand what DPI means. And ask for 300 DPI images as if it meant something on its own.