help with evaluation please

Bug Reports
#1

Hi, I’m evaluating if World Machine will help us on our current project. We are using SMRT/hgt data, baked out in 3dem, and would like to assess if World machine will do more for us than wilbur, and how friendly it is.

Seems simple as there is a project: improving real world terrain. Just snap our image in tweak the settings, and see what we get.

-after bringing in the 512x512, we can’t seem to get a ‘square’ result in the 3d preview to match our image. It’s always rectangular. I’ve created a new file input node, but no luck previewing our low res tile. What am I missing?

-is 512x512 a limitation on input, output or both?

I hoped to finish this evaluation so we could purchase by EOD tomorrow, any help you can be is most appreciated.

Thanks!

talon

#2

Hi there,

The Basic Edition is limited to 513x513 pixels on output; the Standard or Pro editions are unlimited.

If your input file is square (512x512), the following two steps should be it:

  1. Create a File Input device and load your image file into it
  2. Click “Adjust Current Extents” in the file input dialog to make sure the world render extents match the file extents. You should see your entire input file snap into view.

Perhaps a screenshot of what you are seeing would be useful?

#3

ah, ok. Found the size under world extents and resolution, now it’s square. loaded in my image. hit build world (cntrl b) but the first four nodes, including the file input node didn’t build. tried selecting the improved height output, and build world, but same result. Why isn’t it building the world using my input node?

#4

Ok, that’s certainly odd!

If the file input can’t build its usually because of an error reading the file – perhaps file permissions or some such? If there is such an error it is usually reported in the small box around middle-left of the file input device.

If I do the following steps:

  1. Load the example file for real world terrain
  2. Change the file input device to use the png file you posted above
  3. Make the input dimensions square and then hit Adjust Current Extents
  4. Build

I get the following result:

#5

Your image looks exactly like what I have been expecting to be seeing. But for me: no luck.

  1. opened the real world file.

  2. changed the png to mine in the file input node (moved it local in case UNC paths were causing an issue).

  3. The dimensions of the tile are 6.9 km, that’s what I have been putting into the width and height fields in the file input nodes. Do I need to un-check preserve aspect ratios before I hit Adjust current extents?

Unfortunately, it still failed to build the same nodes as the previous screen shot. Perhaps it is something with the evaluation version?

I’m stumped. I run Maya, Mari, photoshop, Mudbox, zbrush, 3dem, so I am familiar with a variety of workflows. I’ve done a bunch of Structure from Motion, so I am familiar with combining scientific based tools into my workflows, but I’m stumped.

Sorry for the hassle, but I love the finish on what I’m seeing without the file input node (just the raw fractals). It’s exactly what we need, but I need to make it work to confirm, and get the PO. I must be missing some basic step. Are you selecting the output node before building? I’ll watch some more tutorials tonight, Hope we can figure out why the build isn’t working!

I’m running windows 7 pro, dual xeon quad core 2ghz , 96 gb ram, nvidea quadro k5000.

Thanks for your patience and assistance!

#6

ahhhh. file must be local. We dropped it in the user folder, not into a temp location on C. now we are getting results.

#7

I’ve just confirmed that there was actually a regression involving network paths in the last version. I will put this in the bugfix list for 2.3.3 – you should be able to load network paths just fine.

#8

Cool, thanks a bunch. Just got hung up on the first chunk not building. Checking out some tuts to improve the results, so I can get a PO. Really appreciate the help and very excited about the tool.

#9

Here is where I’m at:

I created a new project. I imported a crop from our data that is equal to 2.4 km, 2048 resolution. On import it looks correct, but it’s set to 8 km. I set the height and width to 2.4 km. then I go adjust the maximum elevation to get it in the right proportion. So far so good.

I’m seeing very little in the way of previewing the changes adding erosion, or perlin to the existing terrain. Is this because of the 512 limitation of the trial?

I was getting great detailed results yesterday off of the perlin/ erosion set up with no DEM file to start with, why am I not seeing similar results on the the dem file?

#10

The 512 limitation should not be causing any issues.

A couple things to double check:

  1. Are you viewing the right device? If the erosion, etc devices are not selected, you won’t see their effects
  2. The default erosive effect is subtle. Try using a preset in the erosion device like “Classic WM + Power” and see if you don’t see a difference

If all else fails, post a screen capture of how your network is setup!

#11

So I am having trouble getting the amount of detail I want into the terrain still, and understanding the effects of world scale. my dem represents 2.4 km, that I uploaded above. This is our test site.

Ideally I’d like to introduce fine erosion and surface details, like the image I am posting now, but I’m not having much luck using the improved terrain example. Ive went through trying to tweak the settings, and am using a 2k version of the dem.

Any suggestions for cranking up the amount of detail in this type of situation?

I’m completely impressed by the fully generated stuff I have seen, but still in the dark on how tile and scene scale effect the DEM image. I have tweaked settings, unhooked rehooked the nodes to better understand who is doing what, but still coming up short of the results I need.

A tutorial that included the amazing results we are seeing in the sample images but using a dem starting point would be very helpful.

Thanks again!

talon

#12

note the image above was produced with my tweaks on the improved terrain example, with the dem removed…

#13

You have several choices in how to amp up the detail added. Bear in mind that the device parameters were set to work for that particular file, so changing the file will likely require some tweaking and artistic judgement about where you want to go.

So, working again from the improving terrain example file, my process would go like:

  1. Change the file input to your dem and set the scale as needed

  2. Build just the file input node and take a look at its output. Does it have any interpolation artifacts that need fixing? If not, you can probably remove or reduce the blur device in the “mask making” section.
    b) if you see obvious interpolation artifacts, turn up the blur device setting until they disappear.

  3. Select the clamp device right after the Perlin Noise and build-to it. View its output. Do you have a nice grainy surface texture? if not, the noise function needs to be turned up:
    Note that what the perlin noise device is doing is selectively adding small scale noise to its input. Where the noise is added is controlled by the slope selector plugged into the persistence guide.

Inspect the output of the slope selector. Does the mask go to white in some areas or is it just generally dark? Only the white mask areas will be selected for additional detailing, so a good next step is to change the areas the selector is choosing for noise output. Open up the slope selector and adjust its range until you see that you’ve selected as much of the terrain as you would like, then build-to the perlin and see if you’re getting the detail you need.

  1. Erosion works best adding additional detail when you start with a nicely rough surface, so the above step 3 is pretty important. But you can also always play with the erosion device as well to see about a stronger effect.

Here’s an example of the difference step #3 can make: Again using the input dem you posted earlier, here’s a view of the output with the perlin noise device bypassed, and then with it on. Notice how the erosion has much more to work with in the second case!

#14

Thanks for the pointers. I spent some time going through and making adjustments and evaluating them, I just wasn’t getting as good a result as I was if there was no DEM in the process. I’ll take some time today and work with it some more!

#15