<h1 id="depth-maps-du-jour">Depth Maps du Jour</h1><h2 id="laconic:">Laconic:</h2>Depth Maps du Jour is a collection of various images from the abstract to objective. While they were created with ControlNet Depth in mind, the fact that they're simply image files allows them to be used for near whatever purpose might come to mind.<h2 id="loquacious:">Loquacious:</h2>With the depth maps I've uploaded here, you get exactly what it says on the tin - X amount of images that are group in a broad category for your depth-mapping pleasure. I usually start with Pixel Perfect and a Control Weight of 0.75 and toy around from there, but I tend to be looking for unexpected results. Also, I've been making use of the <a target="_blank" rel="ugc" href="https://github.com/DepthAnything/Depth-Anything-V2">Depth Anything V2</a> preprocessor which is fantastic if you want precision in your ControlNet Depth.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b6a32358-b45c-4b3d-84d0-e8e197cf38bf/width=525/b6a32358-b45c-4b3d-84d0-e8e197cf38bf.jpeg" />That said, all the preprocessors and models are equally valid - it's really a matter of what you're looking to do with it. As for packages of images thus far:3,652 Abstract ImagesI haven't personally gotten around to using each and every one of the 3,652 images and, truthfully, there might be a few less because I got rid of doubles, but I had an original count and I'd be damned if I started over. That said, along with the (approximately) 3,652 images, there's also a negative image in a different folder. I use these kind of images to toy with compositions with cool results, IMHO.2,947 Poses &amp; ScenesAs I was doing something of a clearinghouse of old generated images, I thought that just outright deleting everything without making something useful would be a waste - so, I went ahead and converted a good chunk of what I thought was interesting enough to use as a potential pose/depth map and here you see the results. Be sure to lower the weight of ControlNet if you're looking to use them to get more of an approximate pose rather than exactly as seen.

Files with SHA256: 0eb7ebef7825da7b1bcad3243165090034677cc9e39033e6286949c250cee32d