Hey everyone, so I’ve been asked by a ton of people how I create my height maps so I decided to put out this tutorial which hopefully will help a lot of you out. Now there are programs out there that can do this for you but most are paid and/or are limited to what you can do. With Blender though, it’s completely free and very easy to setup. I hope this helps some of you out and I’ll see you all in the next one! 👍🏻
Great tutorial, just a little tip : Once you have selected your camera, if you want to place it at the center of the scene and orient it looking down, you could press alt+g (that will reset it's position) and then alt+r (that will reset it's rotation), i find this more convienient than having to select the fields and set them to zero, but it's just a personal taste, your method is as valid as this.
Great tutorial, thanks! I think it’s important to mention that the Z option in the viewport settings is not related to the fact that you have set the camera on the Z axis. The Z option stands for Z-buffer I think, meant as distance from the camera.
There's a new tool in Blender called Depth Map Batch that generates depth maps for images in a batch process and it's run entirely local on your machine without using APIs and no web services. It's based on a directory of images or a video file.
Hello, and thank you for the video. I have followed your process and the depth map looks good. however, the contrast is too pronounced for my application. i feel that blender makes the gradient range from light to dark too wide. i would LOVE to be able to control that range. When the range is wide it distorts the image (for example the nose appears way out in front of the face. or the ear appears very far away from the front of the face. i’ve had some luck with avoiding these contrast-related distortions with having placing a non-essential object in front of and to the side of my main subject (for example a sphere). doing depth maps in this way…the contrastiness starts with my non-essential object and by the time the depth is portrayed on my essential main objects it has “mellowed” out and so the depth is more accurately portrayed throughout the rest of my essential objects. any ideas on how to control the range of the gradient?
quick tip! you don't need to hold down shit when you select multiple options to change at one time. For example if you want to change the rotation of an object about every axis, you can just click and drag the mouse across the three axis and it will select them all.
Is there any way to render depth map for a model with color texture? For example, an anime model usually use color texure for pupils, if we directly render the depth map from 3d model, the pupils will not show up.
Awesome video, had a question regarding AI images used for relief ive used a depth map docs website and noticed it creates a angled image that isnt flat on the plane could this fix this when adding image that is already greyscaled I tried cnc them but come out tilted on the z axis causing issues with bigger depth on one side which is more cycle time?
AI images aren’t good for relief maps. You end up having to remodel them which kinda defeats the purpose. You could adjust the angle but I typically avoid them unless it’s for a reference image.
@@rockwalldesign you still need to model the design even from a photo. There’s software out there that can do it but they all do a pretty terrible job and aren’t accurate. They might seem fine at first but when inspecting them you can easily tell that they aren’t right. Especially with those AI generated ones. Your best option is to model it.
I have a question, so I took an image, a png file, extruded it, and made it 2.5D so to speak, when I went through these steps on your video, instead of getting this height map, I ended up with basically a black and white version of the image. I've tried removing the png from the material, I've tried saving the mesh I made as an stl, I even tried the stl2png converter but some how my png file is still married to the model mesh and all attempts at the height conversion end up looking like a black and white de saturation of the original file, what am I doing wrong? I had found it amazing that I could "3d-ise" any image, but now I can't make the height map goal unless I figure out how to get these programs to forget about the png base and make the dang mesh the basis for the compositor.
@@michaeltolomei6623 I’d have to see it and you can email me if you want but if you extruded the design straight up, the height map is only going to be black and white as there is no other shades. Try duplicating the design but making 1 of them slightly lower than the other and moving it over by a few mm. That would give it more dimension to check and see. There should be a 3rd level of grey in there if done correctly.
@@ArtByAdrock unfortunately its a bit of a private project, but on a similar note and question, I know that I have seen you do a lot with casting metal as well. And since you are quick to respond, I have a simple question about casting. I have played with sand casting, found some frustrating results, I was wondering, have you played with permanent molds much? I had a thought about making a permanent 2 piece steel mold, and that way I could save on my laser, instead of making lots of engravings and generating lots of dust, just making depth by way of mold and cast instead. use the laser to cut in to steel plates, and make a reproducible result from scrap metal such as brass or aluminum. what do you think? is it worth trying? I know ti would probably make for less need to post process the piece if it casts smooth enough. That's another issue with relying solely on the laser, there's a lot of polishing needed after. I don't necessarily need rounded depth if It is possible to make a steel negative, dremels and trial and error can mimic rounded depth and can be added as the mold is developed.
Is it possible to make a height map from a 3D image? From an AI 3D image using this method This was a great tutorial!! I made my 1st height map today using blender for the 1st time. Thank you! ?
When you say 3D image, is it a stereo image with left and right images for stereo viewing? or something else? There are ways to take the two images and derive a 3D model from them. There is software out there to do it, but I believe it costs money. Do some digging because you may find a free solution, maybe even a solution that gives you a depth map. I know it's out there because I've experimented with it, I just don't have that information handy, but search and you'll find it.
I'm confused, you already have a STL of the monkey, a depth map isnt necessary, you can easily adjust the depth of field to flatten to a flatter monkey. Your intro seemed to indicate converting a dog photo to depth map, would have been more beneficial, I can create 3Dcsrves from that.
It’s really not that bad if you play around with it for a bit but I can see how people can easily get lost in the software. If you’re looking to learn blender I would recommend checking out @RyanKingArt as he puts out some great tutorials for both beginners and experienced. 👍🏻
Hey everyone, so I’ve been asked by a ton of people how I create my height maps so I decided to put out this tutorial which hopefully will help a lot of you out. Now there are programs out there that can do this for you but most are paid and/or are limited to what you can do. With Blender though, it’s completely free and very easy to setup. I hope this helps some of you out and I’ll see you all in the next one! 👍🏻
FINALLY, a comprehensive tutorial on how to create a depth map!🔥👏🤩 Thank you, Thank you, Thank you!
You’re welcome. 👍🏻
I’ve been binge watching videos on this topic and this is probably the best presented way to get consistent and repeatable results. Thank you!
You’re welcome
Great tutorial, just a little tip : Once you have selected your camera, if you want to place it at the center of the scene and orient it looking down, you could press alt+g (that will reset it's position) and then alt+r (that will reset it's rotation), i find this more convienient than having to select the fields and set them to zero, but it's just a personal taste, your method is as valid as this.
I would recommend using Open EXR image format to get the full 32 bit grayscale image. JPG reduces the image to 8 bit and adds JPG artifacting.
This has got to be one of the best tutorials for height maps. If I can laser engrave these, I'll let everyone know.
You saved my life ! Thank you very much for this awesome tutorial.
@@rigel2901 your welcome 👍🏻
This is the best tutorial on zdepth. If you add output file you can automate all the frames saving of the animation
Great tutorial, thanks! I think it’s important to mention that the Z option in the viewport settings is not related to the fact that you have set the camera on the Z axis. The Z option stands for Z-buffer I think, meant as distance from the camera.
There's a new tool in Blender called Depth Map Batch that generates depth maps for images in a batch process and it's run entirely local on your machine without using APIs and no web services. It's based on a directory of images or a video file.
Hello, and thank you for the video. I have followed your process and the depth map looks good. however, the contrast is too pronounced for my application. i feel that blender makes the gradient range from light to dark too wide. i would LOVE to be able to control that range. When the range is wide it distorts the image (for example the nose appears way out in front of the face. or the ear appears very far away from the front of the face.
i’ve had some luck with avoiding these contrast-related distortions with having placing a non-essential object in front of and to the side of my main subject (for example a sphere). doing depth maps in this way…the contrastiness starts with my non-essential object and by the time the depth is portrayed on my essential main objects it has “mellowed” out and so the depth is more accurately portrayed throughout the rest of my essential objects.
any ideas on how to control the range of the gradient?
Thank you for this awesome and detailed video bro! Great work ❤
quick tip! you don't need to hold down shit when you select multiple options to change at one time. For example if you want to change the rotation of an object about every axis, you can just click and drag the mouse across the three axis and it will select them all.
this is a great way to get a 3d depth map for use in stable diffusion control net
Great video man! thanks for the help!
Why does this only work in orthographic view? Any way to get it to work in perspective?
thank you, this is a really very useful video
Thank you so much for sharing this video. You just made my day
You’re welcome! I hope it helps you
Thanks man, that was very helpful.
Is there any way to render depth map for a model with color texture? For example, an anime model usually use color texure for pupils, if we directly render the depth map from 3d model, the pupils will not show up.
No, a depth map only takes the height of a model, the pupils would have to be sculpted and have depth for it to render. Textures will not show up
perfeito amigo.
Hi, is it possible to also provide a version of pure code of generation of depth images as well? I would appreciate that so much!
Hello.
Does depth map = displacement = height map mean the same texture even though the name is different?
This is great, do you have a .blend available with your compositor set up?
nevermind I thought it was more complicated than it is
Well done
Cheers mate!
Awesome video, had a question regarding AI images used for relief ive used a depth map docs website and noticed it creates a angled image that isnt flat on the plane could this fix this when adding image that is already greyscaled
I tried cnc them but come out tilted on the z axis causing issues with bigger depth on one side which is more cycle time?
AI images aren’t good for relief maps. You end up having to remodel them which kinda defeats the purpose. You could adjust the angle but I typically avoid them unless it’s for a reference image.
Hello, please, how do I convert from a regular image to a grayscale image so that I can engrave it with a fiber laser? 3d deep angraving
Is there a tutorial on how to do this from a 2d colour image for portraits
Skip ahead to about 4:20 if you are not a complete blender beginner
Other than the Rick Roll video, this is the most important video on TH-cam. Thank you so much!
what exactly depth map, and why do we need them ?
What size furnace would you recommend for a #8 crucible.
Thank you & keep coming up with your awesome vids.
I always recommend the 10kg devil-forge as they are amazing and haven’t had any issues ever.
Can you make a video about how to make the depth map starting from a photo
@@rockwalldesign you still need to model the design even from a photo. There’s software out there that can do it but they all do a pretty terrible job and aren’t accurate. They might seem fine at first but when inspecting them you can easily tell that they aren’t right. Especially with those AI generated ones. Your best option is to model it.
You can use depth map estimate such Depth Anything V2 you can find in huggingface space. Sorry can't give you link youtube deletes them.
I have a question, so I took an image, a png file, extruded it, and made it 2.5D so to speak, when I went through these steps on your video, instead of getting this height map, I ended up with basically a black and white version of the image. I've tried removing the png from the material, I've tried saving the mesh I made as an stl, I even tried the stl2png converter but some how my png file is still married to the model mesh and all attempts at the height conversion end up looking like a black and white de saturation of the original file, what am I doing wrong? I had found it amazing that I could "3d-ise" any image, but now I can't make the height map goal unless I figure out how to get these programs to forget about the png base and make the dang mesh the basis for the compositor.
@@michaeltolomei6623 I’d have to see it and you can email me if you want but if you extruded the design straight up, the height map is only going to be black and white as there is no other shades. Try duplicating the design but making 1 of them slightly lower than the other and moving it over by a few mm. That would give it more dimension to check and see. There should be a 3rd level of grey in there if done correctly.
@@ArtByAdrock I forgot to add that after extruding, I sculpted the design, it has multiple layers of depth on it.
@@michaeltolomei6623 if you want to email me the design I can take a look at it for you.
@@ArtByAdrock unfortunately its a bit of a private project, but on a similar note and question, I know that I have seen you do a lot with casting metal as well. And since you are quick to respond, I have a simple question about casting. I have played with sand casting, found some frustrating results, I was wondering, have you played with permanent molds much? I had a thought about making a permanent 2 piece steel mold, and that way I could save on my laser, instead of making lots of engravings and generating lots of dust, just making depth by way of mold and cast instead. use the laser to cut in to steel plates, and make a reproducible result from scrap metal such as brass or aluminum. what do you think? is it worth trying? I know ti would probably make for less need to post process the piece if it casts smooth enough. That's another issue with relying solely on the laser, there's a lot of polishing needed after.
I don't necessarily need rounded depth if It is possible to make a steel negative, dremels and trial and error can mimic rounded depth and can be added as the mold is developed.
@michaeltolomei6623 yes I manufacture graphite molds for casting. You could do it with a laser on steel it would just take much much longer.
Nice video, but why are my shades almost all white and have very little depth? I followed all the steps. It looks almost just like a sillouhette
You missed normalize step or the view settings
Is it possible to make a height map from a 3D image? From an AI 3D image using this method
This was a great tutorial!! I made my 1st height map today using blender for the 1st time. Thank you!
?
You would need to convert the AI image to STL format. You need the 3D model first. There are ways to do it but it can be time consuming.
When you say 3D image, is it a stereo image with left and right images for stereo viewing? or something else?
There are ways to take the two images and derive a 3D model from them. There is software out there to do it, but I believe it costs money. Do some digging because you may find a free solution, maybe even a solution that gives you a depth map. I know it's out there because I've experimented with it, I just don't have that information handy, but search and you'll find it.
So you can't just extract the ZBuffer from the GPU, like in Rhino? It takes a second.
nothing appears on my render image :___
when i click render at the end all i get is a solid black outline no depth even though my screen shows an actual black and white image
Did you find the solution?
@@js25121993no I didn’t I gave up for time being thank you
I'm confused, you already have a STL of the monkey, a depth map isnt necessary, you can easily adjust the depth of field to flatten to a flatter monkey.
Your intro seemed to indicate converting a dog photo to depth map, would have been more beneficial, I can create 3Dcsrves from that.
I can’t wrap my head around blender, kudos to you , start a 😮new channel to teach blender, please, 👈I asked nicely 😃
It’s really not that bad if you play around with it for a bit but I can see how people can easily get lost in the software. If you’re looking to learn blender I would recommend checking out @RyanKingArt as he puts out some great tutorials for both beginners and experienced. 👍🏻
@@ArtByAdrock Thank you , I’ll check him out definitely