Messages

1

Gallery / Re: Look development for "Bizerba"

« on: 2017-06-09, 07:02:27 »

Thanks Maru! Yes, pretty much everything was done in Corona. The participating media in the opening shot is Coronas volume fog, it was really impressive to see how fast the renderer did the job here - in an otherwise unlit scene with only one spot light.
Fog / nebula in all other shots were comped, particles for the "library" (the neo-classical structure) were done in pflow.
During an early development stage I had decided to go with a lot of compositing, and this is what was particularly pleasing for me about Corona: individual passes rendered incredibly fast - and we're talking here a 24 core double Xeon from four years ago plus two i7 slaves, so in terms of hardware I was rather limited. Overall rendering time per frame was around 10-15 for simple and 20-40 min full HD for complex scenes with a lot of difficult lighting conditions. I started compositing some shots in fusion​, but yet again switched back to after effects rather quickly, I basically tend to use fusion for tonemapping only - kinda lazy to take care of it during the actual rendering.
I hope I'll find some time to write a more in depth review of the creation process.

2

Gallery / Look development for "Bizerba"

« on: 2017-06-08, 09:56:17 »

Hi everybody,
finally found time to upload my latest solo work.

Not a valid vimeo URL

A lot of work stayed undone (rough and really bad modeling here and there, broken UVs etc.), as there was almost no time to finish it.
Around two years ago I switched my entire production pipeline from Vray to Corona Renderer.
Since then a lot of exciting projects have come and gone, yet I’m still pushing the limits of what the renderer is actually capable of. It’s exciting to see how fast one can iterate through different look development phases.

Comments and critique are much apprecieated, I am very greatful that there is such a great forum community here :D

3

[Max] I need help! / Re: Refraction Exit Color!

« on: 2016-08-08, 09:40:03 »

Hi G-G,
if you look at a photo of a diamond, you will notice the many internal ray refractions / reflections which occur in real life. E.g. take a look at the picture in this article:

https://www.sciencenews.org/article/most-diamonds-share-common-origin-story

Take a closer look at your environment, there's probably a black spot (or the "sky" / environment color), which is causing this. You can try to get reed of it by carefully placing bounce cards in your scene.

4

[Max] Corona Goodies - User Contributions / Re: Corona Mask Manager 1.0 Alpha

« on: 2015-05-24, 11:33:52 »

You're welcome! Be aware of bugs I didn't manage to sort out ;)

5

[Max] Corona Goodies - User Contributions / Corona Mask Manager 1.0 Alpha

« on: 2015-05-24, 01:37:59 »

Hi everybody,
a quick share of a script me and my bro were working on. It had to be a quick one, as I needed it asap in production.
What's the deal? Well, I haven't found an automated renderMask creation script for Corona so far, like AutoMatte for vray for instance (maybe i didn't search enough...)

First try was to create a wire color assign script, which was kindly provided by my brother (as well as the main window and ui programming).
While he was busy with this part, I delved deep into photoshop javascripting, i always find it fascinating to automate something :D
But not being a scripter at all (had just a bit of Java programming some years ago), i put together some pieces and made some really rough rapid prototyping of automatic mask extraction.
Soon I realized that wire color pass was leeding nowhere, as too much mask cleanup in photoshop would have been required.
So i turned to Mono Masks, and added this functionality as well.


wire color extraction in photoshop

wireColor assigment

resulting render elements

1) Maxscript: Corona Mask Manager 1.0 Alpha || Basic functions:

• search scene for all assigned materials and assign 16. predefined wire colors to them. I didn't (and don't) need more for the project. If required you can change the array at anytime to add more colors. Predefined wire colors can be change for individual materials by selecting a material from the list and using the "Pick Wire Color" selector
• create mono masks for all assigned materials, based on selection list includes
• rename mono mask elements according to their correspoding materials (files can be easily stacked later in PS so you know which mat is behind each mask)
• (pending) toolbox with currently a single function to "select objects by mat" (so I didn't have to use material editor for that)

Install: just run the ".ms" file

2) Photoshop javascript: coronaMaskManagerExtractor_1_0_alpha.jsx || Basic functions:
   On the currently opened file perform automated wire color extraction, based on the predefined wire colors in maxscript. At  any time additional colors can be added to the predefined color array.
As I stopped using it and turned to "mono masking" i just provide very few info here:

• the extraction can fail due to resolution for very tiny objects (due to fixed color range threshold inside the script). The solution is to change the threshold value;
• in my tests saving a ".tif" or any other non-HDR file out of corona VFB resulted in wrong wire color values (there is a slight shift). Due to that the script fails in most cases. Solution: save an "EXR"-wire color pass and conver to 8 bit in photoshop
• i had better experience with "non-AA" wire color passes while refining the mask edges after the extraction process. Sidenote: "refine edge" functions in photoshop works differently from standard photoshop "smooth", "feather" etc. so use this if you want to feather and clean up the wire color masks.

Install: drop both files from the "zip" archive in the same folder, then in photoshop open up an image you need the masks to be extracted on. Go to "File > Scripts > Browse..." and select the file "coronaMaskManagerExtractor_1_0_alpha.jsx"
 
Overall notice: I had to work quickly and very rough, so I took a lot of little snippets from the internet. So expect a lot of bugs, as there are probably a lot of situations for the scripts to fail.
The major credits are listed in the header, if somebody feels his / her name is missing let me know. The script is a prototyped tool, I provide it as is, hope somebody can make use of it :)  Please take note of know issues in the maxscript header.
Everyone is free to add / modify anything to their liking, I will be happy to see this thing evolve in time. My current road map is listed in the header as well, provided i get some spare time ;)

Download the files here:
http://www.mediafire.com/download/c278ztimq9u36b4/CMM_1_0_alpha.ms
http://www.mediafire.com/download/43d6dxd5rmu1yz4/coronaMaskManagerExtractor_1_0_alpha.zip

CheerZ,
G.S.

6

General CG Discussion / Re: STEP conversion fails on most of my SiemensNX data

« on: 2015-05-22, 21:57:09 »

pokoy, man i guess you did save my day .)! ThX a lot again, the suggestion of setting of edgeLength indeed made most of the problematic areas go away. I guess i will have to spend a bit more time reading through the nPower manual to understand the tools - so far I've been working only with the Mesh importer, it worked fine up till now.

Regards,
G.S.

7

General CG Discussion / Re: STEP conversion fails on most of my SiemensNX data

« on: 2015-05-22, 14:23:28 »

pokoy thanks a lot for the tip! I am still on max2014 and I have been using powerMeshes all the time, as the viewport performance is way better on my machine. But then I don't have the controls over the mesh tesselation. I will give it a try, really great tip, thanks a lot again :)!

8

General CG Discussion / Re: STEP conversion fails on most of my SiemensNX data

« on: 2015-05-22, 10:50:38 »

Thanks to everybody for replies!

@ Paul Jones: i didn't provide a comparison here, but I have used Vray and the Showcase renderering engine. Both provide much cleaner results (no ugly small triangles), the only places where there were problems in Vray were due to problems in STEP file itself, after rebuilding it in Moments of Inspiration there were no problems. So Showcase and Vray saved my day...

@lacilaci: I would gladly provide you with data, unfortunately everything is under a strict NDA, i am still strugllin' to get at least so parts for public use. Thanks for the tip with zBrush, I'll give it a try as soon as i have a bit more free time.

PS: Rhino conversion yields best results so far, but is not an option - our company is not buying yet another piece of software, after having spent a lot of money for nPower plugins.

Thanks and CheerZ!
G.S.

9

General CG Discussion / STEP conversion fails on most of my SiemensNX data

« on: 2015-05-21, 22:54:24 »

Hi everybody!

I thought I'll post my latest tests in a new thread, to start of fresh. Yet pickin' up our discussion here:
https://forum.corona-renderer.com/index.php/topic,7538.0.html

Over the last two month I have been dealing a lot with new CAD data provided by the client, i struggled day and night and tried a bunch of the major CAD translation packages out there (well, DeltaGen is yet to be tested ;).

To be more specific:
- Maya
- 3dsMax directConnect
- 3dsMax nPower Translators
- VRED Pro
- Showcase
- SAP Visual Author
- Transmagic (to check for any CAD file problems)

The original data was provided through SiemensNX. I yet have to ask for any kind of polygonal data format (like JT) to be provided directly. This way I will be able to test wether tesselating the modell natively without STEP conversion yields any better results...
I can confirm what Ondrej said before: the tiny triangles are the issue. After importing STEP as a rough mesh some of the areas cleaned up. Yet this is not an option for hiRes obviously.

Besides of that I tried using 3DCoat and Mudbox to retopologize the data into quads without luck. There is a new paper for a nurbsToQuad workflow, but it has to be implemented by someone:
http://www.sciencedirect.com/science/article/pii/S0167839614000624
In the screenshots I have collected some results of my testing. You can clearly see that Maya produced clean "unbroken" mesh in places, where there was a solid Nurbs > Quad Mesh conversion. To clean up all the surfaces Maya produces would still somehow defeat the purpose - i do require a simple translator solution for everyday work.





The errors are highly resolution dependent - you might not see the black lines appear on low res, but since most of us are doing high res in production...
I merely wanted to point out the problems I've been having with CAD data for the last two month, I understand that Ondrej is working hard on many different things.
Nevertheless to my mind CAD data conversion / translation / teselation (call it whatever you want ;) ) shouldn't be last priority. I believe there are a lot of people out there like me willing to use Corona for any type of industrial viz.
As for me, unfortunately up till now I had to stop using Corona for CAD - every model turns out to be a pig in a bag. The only option left is to do clay rendering turntables over night (or days!) just to see if any broken normals appear.

10

General CG Discussion / Re: .step files into 3d Max HEEEEEELP

« on: 2015-03-24, 14:51:08 »

Hi GestaltDesign,

thanks for the info, and for the link to your thread :)!
I guess I'll just stick around with the Alpha 0.6 for this particular project, and hope that Ondra will be able to resolve the issue in future versions. Up till now neither do I process the STEP / IGES files myself, nor do i have any influence at what kind of data is being delivered, very unfortunate. But good to know the issue is known!

Thanks a lot again, and thank you Ondra for doing the incredible job out there :)

CheerZzz,
G.S.

11

General CG Discussion / Re: .step files into 3d Max HEEEEEELP

« on: 2015-03-24, 00:53:56 »

Indeed, MOI has a very exotic UI ;) But when you get used to it, you soon realize that you can get incredible results in the shortest amount of time. I too use it in everyday production, mostly as a chamfering / filleting tool.

The one particular issue with it, is when your input files get too large - like 100 mb or so. Then it takes a significant amount of time to open them, although eventually they open up successfully. Never tried rhino though, a lot of people told me that it's a great tool.

Now to the normals issue. For me the same thing, NDA, can't show too much here unfortunately.
But i digged up one comparison. On the left is Alpha 0.72, with the broken normals issue. 1.00.01 produced the same result. Yesterday we finally bought our licenses, I will investigate further. Still, it's definetely a STEP-file issue, i rendered other PowerTranslator imports with Alpha 0.72 without any issues. Litttle spoiler: Vray renders the geometry without any issues ;). I am currently migrating my whole CAD animation workflow to Corona, it simply rocks - unbeatable in terms of tweaking time <> coffee breaks :)

CheerZ,
G.S.

12

General CG Discussion / Re: .step files into 3d Max HEEEEEELP

« on: 2015-03-21, 14:08:18 »

Hi Mr.Schorsch,

I have to deal with CAD-data a lot everyday, in Max the body objects can get quite buggy though.
Then truth is (as far as I know) the Body Objects plugin implemented in Max is a cut-down version of the PowerNurbs package.
After having bought the full one i didn't have any issues in Max while importing STEP or IGES Data (although depending on your package you might need to buy a special version of PowerNurbs, e.g Jupitor Translator [JT]is not included in the PowerTranslators Pro).
There are 30-days trial versions on their page, if you find time check it out! Up till now this was the most ideal way to handle CAD data in Max, without going through the pain of data conversion.
As far it goes for Corona, it works perfectly with PowerTranslators Pro, i have done a lot of tests. All but one produced nice clean renderings with no normal problems. The one time where it went wrong, it was obviously due to bad Siemens NX > STEP conversion. Nevertheless this case is very interesting, because the same geometry rendered clean in Alpha 0.6, in 1.0 it came out broken. I will dig up some screenshots on monday and post them here for comparison, perhaps you guys know the cause.

CheerZ,

G.S.