suggestion for LandscapeTerrain: "assetize" detail masks parameters

[romain.janil]

LandscapeTerrain detail masks parameters are part of the LandscapeTerrain Object and are not external assets as their detail Materials. So their definition belongs to the .world. In an iterative process every time sandworm tools generates an ObjectLandscapeTerrain all the Mask Thresold, contrast, by texel size etc ,are lost.I

n fact all masks parameters (not material) associated to a detail. Besides material definitions, these settings are really important to setup and atm only way I found to apply settings from one sandworm generation to another is to backup .world and edit xml directly... 

Details should be "savable" nodes and so reusable imho.

[bmyagkov]

14 minutes ago, romain.janil said:

Details should be "savable" nodes and so reusable imho.

Got your point, we'll discuss what we can do with this. Thanks for the suggestion!

Please, keep share your experience with us if you have more to say.

 

[romain.janil]

More on that topic: If you "nodify" (make a node reference out of the landscapceTerrain) ObjecLandscapeTerrain then it's possible to regenerate without loosing current masks parameters, otherwise masks are overwritten with default settings. This works only if settings are the same with masks, ie : update albedo or height.

Other insight, if source data is updated or changed, but with same name, cache isn't updated: it would be nice to add "modified date " or so , so that iteration over one particular data (height, alb, mask, vector) with consistent naming would possible without having to clear ALL cache data.

 

[bmyagkov]

Hello!

Quote

More on that topic: If you "nodify" (make a node reference out of the landscapceTerrain) ObjecLandscapeTerrain then it's possible to regenerate without loosing current masks parameters, otherwise masks are overwritten with default settings. This works only if settings are the same with masks, ie : update albedo or height.

Well, nice find but to me it's sounds like a trick which is not much user friendly and I would probably not recommend it for everyone since as you mention above this will only works if the settings match. No good.

We have planned to solve it on Editor side by adding the ability to copy parameters, as the same way it is done for object transformations here:

Unfortunately, I do not promise that this feature will be available in 2.15, since now we are already busy on pre-release stage preparation

 

Quote

Other insight, if source data is updated or changed, but with same name, cache isn't updated: it would be nice to add "modified date " or so , so that iteration over one particular data (height, alb, mask, vector) with consistent naming would possible without having to clear ALL cache data.

Actually, this is already solved in 2.14, when generation process is started the most recent file will be used from disk, but the problem is still remains with the preview image, it will be shown old result used from cache since we can't track changed files all the time. To update preview image you need to restart Sandworm project or manually click on "generate preview" on file from menu by doing right mouse button click

Thanks!

[romain.janil]

Hi, ok yes it's a trick I found, just somehow a workaround, not fully reliable, just wanted to share the discover :) Otherwise yes copy/paste parameters was also what I add in mind, but question: Does this works between 2 open editors? 

As for second usecase, it's odd you say it's solved because I've encountered the problem (ie: change content of some source data, regenerate, and result was still old previous data, i add to rename source file, delete source in sandworm, and add the new one so generation would update with this new data (it was some elevation data), in latest 2.14.1.1 of course. I'm not talking about the preview. I guess reproducing this "bug" will be hard to do tho.

[bmyagkov]

Quote

Does this works between 2 open editors? 

It should work if initially designed right. Perhaps this might be implemented in some other way too, but anyways we will try to add such functionality in some of future versions.

Quote

As for second usecase, it's odd you say it's solved because I've encountered the problem (ie: change content of some source data, regenerate, and result was still old previous data, i add to rename source file, delete source in sandworm, and add the new one so generation would update with this new data (it was some elevation data), in latest 2.14.1.1 of course. I'm not talking about the preview. I guess reproducing this "bug" will be hard to do tho.

Ok, now this sounds like a bug to me. We will try to reproduce the issue on our side. What kind of data we are speaking of? Is this was Mask I guess? Any chance I can ask you to share sources this time?

[romain.janil]

well no, not masks it was just some regular elevation data, nothing in particular: i had a file (only one for elevation), a geotif grid named say "A.tif" that I regenerated at source level (changed some cell values in a lake), open sandworm, deleted the elevation layer, (just in case), re-add this A source file, refreshed preview and hit generate: result was same as before. I had to renamed the file in "B.tif" for the generation to be updated... weird. If this happens again I'll share the data (~250mo rar)

[bmyagkov]

9 minutes ago, romain.janil said:

well no, not masks it was just some regular elevation data, nothing in particular: i had a file (only one for elevation), a geotif grid named say "A.tif" that I regenerated at source level (changed some cell values in a lake), open sandworm, deleted the elevation layer, (just in case), re-add this A source file, refreshed preview and hit generate: result was same as before. I had to renamed the file in "B.tif" for the generation to be updated... weird. If this happens again I'll share the data (~250mo rar)

Thanks for the details, we will look into it asap!