Any idea why my terrain is SUPER low resolution and ‘cross hatched’ at a 2k render?
I’m using DEM data for the terrain with a file input, then everything else, like Advanced Perlin and Erosion, is matched to that of a tutorial I’m following.
The resolution of the build is not the problem, the artefacts are coming from the DEM file which can be relatively low-res as they can be many metres per pixel.
I’m afraid I haven’t used DEMs for some years so can’t remember how to solve the problem short of smoothing the terrain (blur) and then adding back detail by mixing with a fine-grained perln fractal?.
There was a good tutorial on importing and using DEMs (Geekatplay Studios?)
**I think this was the one I was thinking of:-
A good DEM should appear potentially low detail, but still smooth. This looks more like the effects of a subpar format (such as an 8 bit image format) used to store the data. What is the source of this DEM? If you got it from something like “Terrain Party” throw it away, that site is not great. The USGS offers worldwide (and especially US) DEM data freely in high resolution and multiple formats. I’m not sure which World Machine supports, but hopefully it can handle GeoTIFF (32 bit TIFF).
- Oshyan
I tried both and USGS’s scan for this lake was WAAAAY poorer than Terrain Party, yet the Mount Fuji scan was better. I guess it’s hit and miss.
I got it looking somewhat okay though, it was a problem with me upscaling the resolution in Photoshop, when I should have kept it native and used tools in World Machine to improve it.
Terrain Party’s data comes primarily from the USGS, so there’s no way it’s better than USGS. 
There are multiple data types/sources/output formats at the USGS, it can be a little confusing to understand at first. Just go for the lowest “arc minute/second” data or, alternatively, the smallest meter spacing (e.g. 10m, 3m, etc.).
- Oshyan
When I use Terrain Party I get a texture over 1k, if I use USGS the texture is 583 x 583. Maybe you can try and figure out what I’m doing wrong? I’ve tried all the options. The area is ‘Lake Ashi’. They also look TOTALLY different in Photoshop.
Well, the nature of DEM data is that it can be a bit confusing and it takes some reading and experimentation to truly get the best data and the best results. Terrain Party has made a strong effort to remove much of that confusion, but you can still see the uncertainty is there because they provide multiple files with each download.
Currently there are two global data sets that represent the best quality we have world-wide (in the US there is higher quality data available for free, but worldwide it’s lower resolution). Those data sets are SRTM, which used to be only 90 meters per pixel but a new version is now about 30m/pixel (1 arc second) which is called “SRTM 1”, and then ASTER GDEM, which is also 30m/pixel. Generally SRTM is considered more accurate than ASTER, but sometimes you have to download and look at both to see as in some cases one data set looks better than the other for various reasons. Here’s a discussion with some more info:
https://www.researchgate.net/post/Which_DEM_data_is_better_for_elevation_mapping_ASTER_or_SRTM_satellite_data
Fortunately both are available from the USGS Earth Explorer site, so it’s not really that hard to test both.
There is also another newer data set that looks promising called ALOS which may be worth looking at: http://www.eorc.jaxa.jp/ALOS/en/aw3d30/
I tested your area of interest and found some possible detail advantages, however there were some severe spikes of bad data right on the lake edge, so unless you have the means to merge data easily, or clean it up in some other way, the minor additional detail probably isn’t worth it. I am hopeful this data set will be improved over time and may ultimately be better than ASTER and SRTM1 in the long run.
Anyway as far as I can see Terrain Party only offers the ASTER data in 30m/pixel, so that’s a disadvantage, however I will say that this is better than what they used to offer, so it’s clear they’re working to improve on a consistent basis which is great. I’m not a big fan of PNG as a DEM data format for various reasons, but I’ll admit it’s more accessible and widely supported, so I understand the decision to use it. Now that they offer ASTER and a merged product based on other sources (except SRTM1) that can make up for some of the issues in ASTER, it’s a decent source, and I’ll have to apologize for not re-checking what they are currently offering before leveling criticism earlier. The last time I tested their data was probably 3-6 months ago and it was more limited and lower quality.
As far as the raw ASTER and SRTM1 data itself, it isn’t going to look right in Photoshop. The only reason the Terrain Party stuff does is it uses 16 bit PNG, and PNG is a more widely supported format in image editors. There are good reasons to not use the PNG format however, including greater accuracy and important meta data like georeferencing info (which World Machine doesn’t read at present, but is useful in other applications like Terragen).
So in the end Terrain Party is overall a decent option. It’s great for ease of use, it is absolutely the easiest approach to get the data you want. It does not guarantee highest quality, so if quality is a concern, go with the USGS ASTER or SRTM source, but currently the difference is not huge.
If you want to directly access either SRTM1 or ASTER data you can find both here: https://earthexplorer.usgs.gov/
You’ll need an account to download it, but it’s free. The ASTER data also has particular usage restrictions, but is still free to download.
Your area of interest appears to be near a tile boundary, but the tile I got was 3601x3601 in both ASTER and SRTM 1. It’s hard to say but I think the SRTM 1 data is slightly more accurate/detailed. You’ll need 2 tiles to get the whole lake since it’s right at a tile boundary.
- Oshyan
