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 - Ricky Johnson

Pages: [1] 2 3 ... 8
1
Quote

Hi Ricky,

Are you able to install the latest version of Rail Clone? In v6 we updated to their latest API so instancing should be working fine.

We do have a test for this which is passing, so can I ask you to send over your scene and we will take a look at it?

Thanks,

Rowan

Hi,

Yes, I've made the update to Railclone to version 4.2.0 now.

This doesn't seem to have any effect on the original point I raised but I can see that Railclone's internal material ID randomising node is now working with Corona. So, this does help in providing an alternative route to achieve similar results in future.

Just to be clear about the original point though - using a 'mesh element' Multimap in a Railclone object still no longer has the desired effect on Railclone instances. I've attached examples again - the only change here is different daily build versions of Corona.

Are you getting this to work any differently with the latest build? I wasn't sure if you were talking about getting the Railclone material node working in your own tests or whether you're referring to the 'mesh element' behaviour in Multimap.

I could send this scene but it's just a basic set up made from scratch to illustrate this point. I've not found this issue to be scene dependent. Let me know if it's needed though.

2
Quote
Hi,

A couple of things changed around then.. if all slots have the frequency set to 0 then all slots are picked equally based on random amount.

Also the seed randomisation changed, maybe you can try to change the seed in the multimap and you will get a different mix of colours.

Let me know if this helps.

Rowan

Hi Rowan,

Thanks but I get the same result if Frequency for all slots is set to 0 or 1 (I've attached a simple example using Railclone Library style).
The seed in the multimap does change the slot being picked from but all of the instanced elements in the Railclone object are still treated as the same mesh element and get assigned this same slot.

As Nik suggests - Disabling the Instancing Engine in the Railclone object - gets us back to the result from before. But this does mean losing the savings on RAM usage that Railclone offers.
Up until this latest Daily you could get randomised slots for all elements (instanced or not) with the advantages of the Railclone Instancing Engine as well.

Anyway, if it's unavoidable due to changes overall, then that's that.
Similar results could be achieved with material ID's with a bit more effort (although would be great if Railclone internal Material ID randomisation worked with Corona here also).
Just a shame to lose that simple set up and ease of compatibility with older scenes that might use it.

3
It seems that the behaviour of CoronaMultimap applied to a Railclone object has changed in the latest Daily Build.

Previously Mode 'Mesh Element' in MultiMap would assign a different Colour/Map to every single object generated by Railclone. Using the latest Daily - Instances in the Railclone object look as though they're grouped together and assigned the same Colour/Map.

I've tested this against the Daily Build from 28th April - everything is normal as before in that version.

This is on 3ds Max 2016, Railclone Pro 3.3.1

4
General CG Discussion / Re: Ultra High Resolution Wood Scans
« on: 2020-05-15, 10:15:01 »
Some of the Arroway Wood veneer scans get to about 15k along the grain. They also offer a good selection.
https://www.arroway-textures.ch/en

I'll be interested to see what suggestions you get on this.
Arroway made these sets quite some time ago (Although they have reworked the maps and updated them since) but it still looks like the most comprehensive collection of 'fine' veneers.
With all of the newer texture providers on the market now, nobody else seems to have attempted their own version of something comparable.

5
Hey Ricky,

I'm also a freelancer doing animations. I was used to rebus and it worked quite well then I tried ranchcomputing and I was not disappointed or missing anything: https://www.ranchcomputing.com/fr . It works nicely, the launching process is easy and it's cheaper.

Thanks Fluss,

I looked at 2 alternatives, when this happened with Rebus a few weeks back. Ranch was one of them and GarageFarm the other. They both seemed pretty good.
I used GarageFarm in the end for the project but it was a very close call. I think it was just upload speed to GarageFarm seemed quicker for me at the time.

So yes, I agree there are good alternatives to Rebusfarm out there (without the pass limitation).

6
I just checked back in on Rebusfarm support about this. I had a response on the 19th:

They said that the pass limit was introduced because they were experiencing many complaints from users who'd set passes to 0 and were using noise limit only - resulting in lots of claims for refunds.
After observing many Corona jobs on the farm and running some tests internally they came up with the figure of 400 as a limit that could cope with any situation.

They did suggest that they were open to increasing the maximum amount and it wouldn't be a problem if there were users who felt this was necessary in some cases.

I've just replied to say that I felt it was too low for some cases and that ideally 0 passes should still be an option, perhaps with a big warning note about potential render times if set to 0.

Also suggested that they start a thread on the forum here to assess support for this request.

Don't know if you got anywhere with talking to them about this Maru?

7
Do you consider instead of throwing money to rebus to hire someone who will optimize your scene for 100-200pass max? I never render 200+ but i know this was also too much with wrong setup.

I am interested to know if 400 passes is typically enough for every situation for most users, so thanks that's good insight.
For my own part I'm generally much happier to keep a scene complicated and make up the extra cost associated with that choice myself.
I'm a freelancer who works in isolation so my main source for comparison is info from this forum - I will have a look at suggestions for optimisation.

An issue I was having previously with this scene was that individual test frames would look okay in terms of noise at a lower level of passes but animations revealed swimming noise in GI in some areas, glossy reflections in others.
Therefore, the testing itself gets more time consuming, expensive and speculative (having to test sequences of animations in motion rather than being able to rely on individual frames).
So, yes, 600 may seem high but I am at least confident that I had something that worked and I'm not going to be spending money on final renders that I can't use.

Anyhow, this choice on the users part is a separate issue from the fact that some period of notice to the changes in Rebusfarm's capacity would have been professionally sound.
They do invite customers to buy credit for the service based on the principle of a few test renders working to the customer's satisfaction.
If renders can run at 600 passes one day and then cannot do so the next day without prior notice this is not great practice.
I've yet to hear back from them whether it's possible to temporarily bypass the limit though. Maybe they'll offer a solution.

8
Just a notice for people using Corona with Rebusfarm:
(This is not official. I'm just a customer who has realised this by trial and error, followed by contacting Rebusfarm support today)

As of an update to the Rebusfarm plugin today, there is a limit on the maximum number of passes you can specify when rendering non-distributed*. Maximum is now set at 400 passes.

This seems to be by design as part of the update. I've asked for a bit more information from Rebusfarm support and am waiting to hear back.
I don't currently know if it's possible for customers currently in the middle of a project to roll back to an earlier version of the plugin without the limitation so they can finish their work with consistency. This is one of the questions I've asked about.

*I don't think this limit is imposed on distributed renders - scene checking with distributed render does not throw up the same error that occurs when trying to exceed 400 passes non-distributed. That's all I know.

I don't know if this would be of concern to anyone here in terms of them providing the most viable service?
I'm currently preparing a flythrough animation of a complicated interior I've tested to the point that I feel 600 passes are necessary. Maybe that's too unusual to cater for. I'm awaiting to hear of a reason for the change anyway.

9
There is supposed to be a technically correct method that requires treating each channel separately.
Lots of guides are available for different applications - Like this one: https://www.moddb.com/groups/udk-developers-group/tutorials/photoshop-combining-normal-maps
I don't know if the same set up could be replicated within a material in 3ds max - it would be complicated. The blend mode overlay always looked okay to me for a simpler set up.

10
General CG Discussion / Re: Eucalyptus trees
« on: 2018-05-16, 17:33:43 »
There's an Australian based site here, might be worth a look also:

https://www.3dmk.com.au/

I don't know if they have Eucalyptus trees specifically - always looked like decent quality stuff to me though. They sell Grow FX versions of the plants too, usefully.

11
General CG Discussion / Re: Blender new viewport Eevee
« on: 2017-07-28, 11:58:06 »
the debt in 3ds max is probably super duper high.

I reckon that Autodesk's profit margin is super duper higher than the debt in the code. :)

Even given the magnitude of the task, they could surely employ the resources to deliver a product worthy of the price we all pay endlessly month after month. Obviously the organisation has become too big to act in a decent human fashion.

It's a shame that it would be so impractical for all subscribers to 'down tools' on 3ds max for a month in a coordinated protest of some kind!

12
Work in Progress/Tests / Re: Dragon portrait
« on: 2017-07-05, 11:00:17 »
It's looking incredibly good! The textures on the horns seems to match the model perfectly and appears instantly plausible (for a dragon). Will be very interested to watch the progression. Inspiring!

13
Those all look quite similar to the way Arroway produce their larger textures to me. You didn't mention if you were willing to pay or wanted them free though.

I think some of their wood textures are available to buy as singles now but most are bundled up in the volumes. Depends what species/grain you're after.

14
are saying i can bake displacement tex into models in this way?

You can in 3ds Max. I'm afraid I don't know anything about C4D but am just guessing that a similar export process may exist that you can use to the same effect.

15
i dont think it would work on c4d coronas, as i already tried proxying. would you suggest baking displaced texture on object completely?

Sorry, I missed that this was a C4D thread. I can't be of too much use then. I don't know if there's an equivalent method for C4D - Isn't there the option to export the scene from C4D in it's entirety?

The 3ds Max method relies on exporting the entire scene (I suppose the typical intention is to go to a standalone). The export process then triggers the displacement calculations and bakes them into the proxies saved.
These proxies can the be loaded into your scene in place of the current objects with displacement.

Pages: [1] 2 3 ... 8