Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - grasshopper

Pages: [1] 2
1
Thanks

2

+++100
I don't remember when was the last time I used a new converter in last Corona versions/  it is absolutely impossible to use it because you will never be even approximately sure of the final result. especially nice with multimaterials from others models libraries/ I just kept the old converter as a script.

Hi Denisgo, Don't suppose you have a copy of that script you can post? or a link where I can get it please.

Thanks

3
I have to say, wither you regard it as a step forward in material creation, physical accuracy or anything else, it is a huge step back in workflow and usability. Converting almost all materials to metals and a load of materials to max physical materials is totally useless and makes it completely unusable.

The old converter was excellent, and to say if you are not including a converter for legacy materials because it is not the way forward it is incredibly arrogant. We need the tools to do our jobs so get it implemented so we can work, and then when you get a working Corona 7/8 converter phase out the old one if you have to.

I have very little interest it converting from corona 6 to 7 or 8, we wouldn't render jobs across 2 different versions. What we NEED is a quick and reliable way to get models from clients and get them in a state where we can start improving them. Not re-texturing them from the start.

Please put your users first not your ideals.

Rant over
thanks :)

4
Thats right Frood, So I am thinking it will be more of a Settings override system. We are working remotely atm as we are isolated. So we set up scenes as we need them locally, sync all scenes back to the office and then am looking for a way to send renders to the farm with the option to change useful settings without having to re-open the files. It's something that might be relevant to a lot of people at the moment.

As you mentioned we are not able to access the max file settings and simply change the existing settings so it relies on knowledge of the scene being rendered but I think it still justifies being a worth while exercise.

One issue I notice when sending to backburner using commandline is that without specifying the camera (and the only way is knowing the camera by name) it is always perspective of ortho even if the camera is specified correctly in the saved file. I can send suspeded and change the view in backburner but it would be great to have a solution for that.

I managed to write a script (very pleased with myself even if it is simple) and it analyses the open scene and outputs a txt list of the cameras in a specified location. The Unity guy can use that list to produce a camera selection drop down.

Cams = for Cam in cameras where (superclassof Cam == camera) collect Cam
if Cams.count > 0 then
out_file = createfile ("C:\\temp\TestCams.txt")
for Cam in Cams do format "%\n" Cam.name to:out_file
close out_file

Any idea how I can I pre run this script prior to sending the render (get camera list button) without opening the max file?

5
Thanks again Frood,

As I said I was sending render by command line through batch script but the process is growing arms and legs now a colleague of mine has got involved and he is familiar with unity and is putting together a UI allowing me to submit jobs and change some options. His interface will basically output a batch file which I can run to send jobs to render without opening them. My colleague isn't familiar with corona or maxscript so trying to work out what changes we can make to render settings. Pass limit was a must but prior to your response I could only change it through render presets which is ideal. To add noise thresholds, changing from still frame to animation and changing post\lightmix setting would be a few other useful settings. The script by Phasma that you pointed to in one of your posts looks very useful but it's going to take me a while to digest all of this and investigate further.

Thanks again for your insight, I fear you may not of heard the last of me on this subject. :)

6
Thanks Frood, I am not an experienced coder so didn't know about that. I'll add a prescript as you have shown and try and get it working. I had reverted to loading a render preset but that isn't ideal as has a lot of changeable values and would need a separate file for every alternative. (and they are encoded so can't be edited ina txt editor)

Is there reference anywhere to other script commands for corona settings?
Also can you run more than one prescript from each bat file?

Thanks again you have opened up a new avenue for investigation.

7
I have recently been playing with submitting jobs through command line and am now trying to create custom bat file to control my renders.

Is there any way to change the pass limit or noise threshold limit or indeed any on the corona settings whist submitting through command line?

Sorry I forgot to note 3dsMax 2018 & Corona 5

Thanks
G

8
Just read the post you linked to and it describes exactly the same behaviour.

Glad it's not just me. A fix would be awesome :)

9
Thats all correct Maru, thanks for looking into it.

I don't want to use a region but completely by accident noticed the difference it makes.

10
Previous post:  https://forum.corona-renderer.com/index.php?topic=27617.0
I am making this post in relation to a previous feature request that I made and found a work around for.
The bases of the request was to use an object in the scene as a render region using a gbuffer ID. The work around uses an object with a material set as transparent so in theory it should just render everything within it’s form. Great for patch rendering animated scenes with cameras and objects moving.

So I did some more testing and have an issue as the render behaves differently when rendering selected ID is set. The render takes long time and renders to a much lower than expected noise threshold.
All attached renders are set to 40 passes, and I would expect it to reach around 2-3% noise. The first render has render selected by Gbuffer ID and the ONLY difference in the second render is that I drew a region tight in around the rendered area. The third render is EXACTLY the same except the region covers the whole frame.

First Render (no region selected ID only)
render time -26 minutes
noise level - 0.96%

Second Render (with tight region)
render time -7 minutes
noise level - 2.63%

Third Render (with full frame region)
render time -26 minutes
noise level - 0.96%

The second image is how I would of expected the first image to also render. In fact I would of expected all 3 to render the same!
Although there is a time saving in set up time it leaves relatively little point in using gbuffer instead of rendering many regions.

I would expect a whole frame to render in about 45 minutes. We have had similar experiences with render time in the past using Obj ID (never been trying it with a transparent object before) but never really investigated it any further other than rendering to a noise threshold which for animation we found flickery.
Why would a region bigger that the object effect the render so much and Is this an expected behaviour? Has anyone else experience similar? and most importantly is there a solution (other than regions)
Thanks for your time.

11
Hi Guys, Thanks for your responses, sorry it has taken me a wee while to look back at this.

Frood, you are right it needs to work on a scene with everything animated. They are heavy complex scenes too hence trying to save time on rendering patches in. I had never tested with the auto region select, in fact its not a feature I have every used before. Would work well but it does need to work on a network render.

I have been testing this for a while and never came up with a satisfactory solution. I was setting refraction to 'thin' thinking it was the same as 1.0 but it would then render nothing. Setting full opacity would render nothing too. I found that setting visibility to 0 in obj props rendered nothing too but now after looking at your suggestions and with further testing I must of been combining that with zeroing the refraction.

What I find now is that unless refraction is turned up from 0 neither opacity or visibility work but as long as refraction is at 0.001 or above it will render the selected objects area so I will progress along this route with a full test animation and see how it works.

I still stand by my feature request though a simple dummy render object in scene would be awesome it could even have a fall off for feathering in in post.


12
Being able to used viewport selection or Object ID as a selection for rendering with the selected object set to invisible to camera would be awesome for patch rendering and fixing animations.

We use xrefs to build up our master scenes so don't have geometry in our render scenes. The ability to drop some geometry into a render scene and use it as a mask for rendering selected on ID would allow to render a very specific areas through an animation. Currently if renderable is unchecked or invisible to camera is set in the properties that object cant be used as a selector. This would allow you to create geometry larger than your target to include shadows and GI, you could link such an object to animated models in your scene and re-render allowing enough render data to feather results into previously rendered footage.

Hope this makes sense. :)

13
This would be an awesome tool and would save a huge amount of time and hassle whilst producing animations. Batch processing is a step in the right directions but an After effects plugin that could produce the same results would be amazing.... Please make it so!!!!! :)

14
[Max] Feature Requests / Re: Batch process to 16bit exr
« on: 2019-06-04, 11:31:42 »
thanks, would be useful.

15
[Max] Feature Requests / Batch process to 16bit exr
« on: 2019-06-03, 12:52:19 »
Hi,

Is there any way when using the batch processes to export to a 16bit exr?

We don't need much adjustment after running the batch processor and the file size is significantly smaller at 16bit.

Thanks

Pages: [1] 2