How do I make a 3 dimensional character out of flat picture?

Question

I have a cartoon character (a lion) that is part of my business logo that I want to put into Blender and animate it somehow. Is there any way I can take a flat picture (JPEG) and transform it into a 3D image and then animate it in Blender?

Answer 1

You may do it using the Transformers add on. It'll transform every flat textured plane into 3d model.

Activate it in User Preferences. Then go to a Tool Shelf and choose a desired transform option.

OKAY I'M KIDDING :), NOW SERIOUSLY:

There's no such way to do it fast. If you manage to invent a tool that may do such thing automatically you'll be probably a millionaire :). While there's a very simple way to transform 3d model into 2D image (using e.g. shapekeys), reverse process is impossible (at least right now :)).

As @Shubol3D said there are no shortcuts. You'll have to master a lot of skills to create your lion model. It's very time consuming process, so if you like to save it I suggest you to go to Blenderartists.org forum and make a paid job offer there. There are a lot of super talented people, who may do it for you.

But if you still want to do it by yourself...

Here are some ultra basic 3d model production steps:

First of all get familiar with this topic: Resources for Blender It's super useful for the blender beginners.

MODELING

You should start from setting up the reference images for modelling:

• How do I set a background image while I am editing?
• How do I scale my reference images correctly within Blender and is there an easier way?
• Modeling with reference image

First add a cube. Use Toggle Quad View option (Ctrl+Alt+Q) to split the screen into quarters and set separate background images for each one. Adjust their positions, so they fit the cube position in Top, Front and Right 3D views.

To make the modeling workflow faster use a Mirror Modifier:

• Is their a way to create half of a model and duplicate other side

• How can I make the mirror modifier work right?

Divide the cube into two identical parts with Ctrl+R and delete one half of it (X). Add a Mirror Modifier to it. Now you may model just one half of the mesh and the other one'll behave symmetrical.

Enter Edit Mode and start modeling basing on the background images you set:

• What techniques are used to begin creating new models?

There are few buttons which are very helpful in modeling: G- for grabing/translating vertices, edges and faces, S- for scaling them, R- for rotating, E for extruding and F for filling the holes. You may also subdivde the desired faces using Ctrl+R shortcut.

Start with extruding the cube to create a basic shape that fits the background drawings. Then subdivide it successfuly to add more and more geometry and details.

Thanks to quad view you may control the shape of the mesh in every view at once. Fit it to the background drawings as accurately as you can.

Finally you may add a Subsurf Modifier to make your model smooth. Go to the Modifiers header, activate it, set the desired subdivision levels and smooth it even more with Smooth shading option in Edit panel of the Tool Shelf.

I think that in your case this tutorial may be a good start: https://www.youtube.com/watch?v=6mlGT6T9bfQ It shows how to create a 3D animal model based on the 2D images.

UNWRAPPING AND TEXTURING

To unwrap your model properly you may need to make seams for it:

• How to properly unwrap my mesh?

• How to Place the Right Seams for UV Unwrapping

Select the edges you want to put the seams on and press Ctrl+E-->Mark Seam. Now select the whole mesh (A) and press U-->Unwrap.

Now you want to put the texture on your model. Though there are many methods to do it, I prefer the texture painting with Stencil method:

• How to make textures in Blender without any flat work with a painting program? (3D painting)

Select your model, set up the shading type to Texture and go to Texture Paint mode. Add a new Paint Slot-->Diffuse Color in a Tools bookmark of a Tool Shelf (T). Add a new texture in the Texture panel on the left, then go to the Texture header on the right, select Brush and open a texture, which will be used as a stencil for painting. Paint the texture and save it.

Now add a material to the object and assign a newly created image texture/textures to it:

• Add an image to an object (Blender Internal Render)
• How are texture maps imported into Blender? (Cycles Render)

Go to the Material header and add a new material (click New button).

IN BLENDER INTERNAL: Go to Texture header to assign a texture to it. Add a new texture, set its type to Image or Movie. In Influence panel check the Color checkbox. Set the viewport shading to Texture to display it on model.

IN CYCLES RENDER: Go to Node Editor, check the Use Nodes checkbox and set up nodes as pictured below.

Basic tutorial showing how to UV unwrap an animal model and apply a texture to it: https://www.youtube.com/watch?v=p4ngVoGIj1Q

RIGGING AND ANIMATION

Now you have to construct the rig (skeleton) and parent it with your model:

• how to rig the body
• How can I automatically generate vertex groups for an armature?

Add a single bone (Shift+A-->Armature-->Single Bone) and extrude it (E) in Edit Mode creating a skeleton. Limbs can be mirrored from one side to other using Ctrl+M command.

When your skeleton is ready you may you may parent your mesh with it. In Object Mode select your model, then your armature and press Ctrl+P-->With Automatic Weights. It'll automatically create specific vertex groups for the specific bones.

Finally you can animate your bones:

• How to make an animation / armature remember poses?

Select a bone in Pose Mode, translate it using R,G or S and press I to keyframe it. Go to the next frame on the Timeline, translate it again and keyframe again. Press Alt+A to play the animation. Adjust your keyframes spacing in Dope Sheet Editor. Control every aspect of your animation using curves in Graph Editor.

Very, very basic tutorial showing how to make the skeleton for your animal and how to animate it: https://www.youtube.com/watch?v=Kr6-_NT_olo

Answer 2

Use Photoscan,and if it didn't turn out like you wanted then you can use the shapes to make it,but I reccomend using MakeHuman or the Manuel Bastioni Laboratory. I'm making a FNaF fan game with human characters,and when I finally found out about this,I used the sliders to make the character exactly like I wanted to. This is your answer. But if it's a lion you're talking about,I highly recommend that you use reference images to sculpt the anatomy of the lion without mane first,then you model it using the sculpture as a base,then use the Subdivision.

The Subdivision Surface modifier will increase the polygon number,I also recommend you to use image textures from the internet to trace the shape and for symmetry,use a mirror modifier for one side,then when you apply it,swap the weights with their opposites. Then you can detach a limb,one by one and parent it to the armature set with Automatic Weights. Then add mane using hair particles and go to the Blender Guru's site to learn how to make realistic fur.

Answer 3

The question is seven years old, and the two answers are almost as old, so it's time for an update. The answer is now:

Yes, there is a way. A very easy way.

I don't know if this is the only way, but I found a very easy way in a YouTube video by Nik Kottmann:

• Go to the free online tool, Monster Mash.
• Import the picture.
• Draw lines around the segments of the object, such as the body, legs, and ears of an animal.
• Click "Inflate" to magically turn the picture into a 3D object.
• Click "Animate" (not to animate the object, but just to turn on the option for the next step).
• Export the object as a glTF (.glb) file.
• Import the glTF file into Blender.
• To see the surface textures from the original picture, change Viewport Shading to "Material Preview".
• Switch to Edit Mode to reveal the vertices of the object and modify the object as desired.

The video runs through all the steps for doing the above. Nik takes a picture of a rhinoceros and turns it into a 3D object in Blender in a few minutes. I ran through his steps with a picture I generated in Leonardo and got the same result.

Pretty amazingly easy!

Answer 4

We're finally here! I wrote in multiple places that the AI tools are extremely limited and can be replaced by a very basic modelling skill. This changes with...

www.meshy.ai

• Click on the green "Get Started" button
• Register new (free) account
• Click on "Image to 3D"

• Drag an image in this field:

• Click the green "✨ Generate" button

Here's my test: I started with a PNG file.

I actually used 10 times (in each dimension) bigger image, but new BSE limitations didn't allow me to upload it. I don't know if having a high resolution image was beneficial... It's a logo of an armadillo. It's flat - monochromatic: blue/transparent. The background and features of the armadillo share a color (transparent). It's abstract - it isn't an actual real photo traced into an illustration, it doesn't have proper shape of an armadillo nor proper perspective. So it's a very tricky input. Here's the result from Meshy:

It successfully differentiated the features from the background (as opposed to making holes in the face), as well as figured the shape just about right. Well, "about" is the keyword here, let's look at the topology:

It looks like playdough shaped by a 6 year old kid. If you thought it is created by a human, you wouldn't be impressed. But because it's created by AI you are impressed... It kind of looks like low poly, but is high poly.

Meshy also allows you to download the file, with one of the formats being .blend!

In the right panel, you have an option to decimate and to convert to quads as well!

Low poly hides the imperfections, however it still looks like a very shoddy job because it's asymmetric:

As a comparison, here's what I modeled manually (before even knowing about Meshy, so it didn't affect my work in any way) in a few hours:

The topo of ears is weird, because at some point I applied subdivision and sculpted, so before taking the screenshot I added a "decimate: unsubdivide" modifier. So the topology could definitely be further improved if a better workflow was used. The coloring was done by casting a shadow from the original image in Cycles, and then baking it into a texture.

It begs the question, however, what if I took the render of my work as an input for Meshy?

The ears are asymmetric by design (they are asymmetric in my model):

It seems the AI Doomsday is still not going to happen tomorrow, but perhaps Meshy is a good tool to use to quickly prepare a base shape + texture for further sculpting. You will end up, however, with a topology that doesn't have some logical connection with your shape and isn't symmetric.