Hello,
I would like to know if anyone have an idea of the workflow or the technique to aim to reproduce the morphing/merging animation of this insane Title Sequence ? I really can’t find how this work.
It’s not as complicated as it looks- it’s objects that appear to be one object because they’re touching, parented to different (probably) empties so they can move apart from each other. You can make the complex object and then split it into pieces with P. The snakes on the bottom left are just an object moving along a path with a Follow Path modifier and an animated offset. The rest of the deforms you can do with shapekeys
1 Like
I think it’s done on several steps, first everything starts to move, like the snakes, the first image is deconstructed.
Then basically , I think they manage to took all these elements, deform them further and form the logo as best as they can. That will probably led to a kind of archiboldo’s result :
Then they have a finished sculpted version of the logo, that starts to appears on top of that.
And by then end it recovers every individual pieces.
It’s not something that complicated, but it takes time. They probably work on that sequences for a few mounts. Also there is always a compositing pass on top of it that helps to fix stuff, or maybe allow better transition from the archimboldo state to the final.
Because it’s very well done, and done in different steps it gets difficult to retro-engineer exactly what happens. But that’s probably something along these lines.
Bear in mind that you see only the final result, they may have done some initial versions that didn’t work so well and improve them further. Sometimes you need a rough pass and then work further on details or fixing stuff even if 70% are working.
It can get simpler if you start thinking in different steps, rather that one technique only that do the trick.
If that’s not enough to get you started , maybe then we can talk about different ways of deforming objects, how to achieve some transitions, animate stuff etc… Feel free to ask more specific questions if needed !
2 Likes
okay I start to kinda see it now !
I’ll experiment some stuffs and I will certainly come back with questions, thanks a lot for your help !
I have a first question, at the end of the animation, what is the process to make it appear like this ? Is it like a dissolve technique that you can do with geometry nodes ?
Hum !! By having a second look I think they added another step. So to sum up the end.
1/ Every pieces goes into an archimboldo state.
2/ Then a fluid like version of the logo starts to recover the archimboldo state.
3/ At compositing they used a clean rendered version of the sculpted logo and make it appear with a fade / dissolve effect, probably using a mask to make it non-uniform.
Step 1/ : we talked about it.
Step 2/ : it’s quite well done, it looks a bit like a fluid/water effect ( more honey than water) but it’s very precise.
If I had to redo that with blender, I’ll mix a couple of modifiers.
- I’ll start with a basic mesh that forms the logo (not the sculpted one) .
- I’ll add a boolean modifier on it, with an sphere as the cutter (the sphere recovers totally the logo. By making the sphere scale from 0.0001 to 1, you’ll have something that works but quite boring.
Not sure of the exact settings here. The idea is that the sphere cuts the logo so at scale 0.001 you barely see the logo, and at scale 1 it’s revealed completely. - At this stage, I’ll try to add a displace modifier on the sphere with a noise as a texture. That noise you can control it with an empty (in the modifier settings) . By changing the size of the empty, and animate it’s location slightly, you’ll get a watery effect on the sphere, that will then influence the boolean cut resulting in a bit more interesting result.
- from there, I’ll try to improve the effect a bit more on the logo. One thing to try, is to add a remesh modifier after the boolean to get an even mesh, and add a displace modifier on top of it, with the same empty driven texture. Of course, it’s worth testing different textures, clouds/noise, voronoi… you can get very different results.
- you can also layer several displaces, with different textures. add a subdiv modifier to get more subdivisions and or smooth the result.
- another thing, if that’s still not enough, you can try to add a shape key (animated) on the logo mesh and add more variations on top of that .
It’s also possible to use only shape keys instead of the boolean modifier.
You may also get crazy with geometry nodes and get interesting results, but that would be probably not needed to get something cool especially if you’re starting with GN, it may slow you down without much benefits.
You can try the modifier version, and redo that with nodes as a basis if you like, then see if you can add more layers of deformation on top of it.
As you guessed, it’s an accumulation of many simple effects that will do something complex enough to trick the brain. So we don’t see just an animated boolean, or an animated displace, but something that looks more complex.
On these kind of effect, the best is to take simple effects and look for variations. A different texture for displace will give you very different visual results. Instead of a sphere as boolean you can try other shapes, maybe a model that you animate with shape keys instead of the scale + a displace like we talked before.
You’ll eventually find a recipe that works, you may also find another way that is very different from what I said.
I’m mainly freestyling / brainstorming on how I’ll approach it before testing, but along the road I may tests other things , or get new ideas as well. It’s a matter of deforming object with more or less control. And layering these deformations to make it look complex.
3/ Finally , the final reveal of the logo.
They probably had a rendered version of the final sculpted logo. And use a compositing application to make it appear. On this part, it’s not that complicated, worse case scenario a cross fade can give something, but obviously something a bit more elaborated will work much better.
If doing something in a compositing application is something that scares you… You can try to take the sculpted logo, and animate it’s transparency. That will give you a cross fade. Now you can take a noise texture, and try to manage to have more or less transparent parts during the fade.
But many things can do.
Another way of doing, you find a cool trick to make the final logo appear, maybe something geometry node based like this : https://www.youtube.com/watch?v=35AM0t-9XCg
And you may even not need the step 2. That may be enough to sell the transition from step 1 to 3.
Sorry to add another layer of confusing information, hopes that helps. The more you’ll get the logic of making stuff appear / disapear in different way, you’ll see that you can customize the recipe a lot here to get interesting results. Once again it’s just about taking quite simple effects, and mix them to create a sense of complexity and trick the mind. By replacing one effect with another ( that you’re most comfortable with) you’ll get interesting results anyway. And once different effects will start to mix you’ll get that feel of complexity that you’re looking for.
Feel free to keep asking question, probably one answer lead to other questions…
It’s also worth trying stuff on a simplified scene, like how to make step 2. Just to get comfortable with the techniques. And latter incorporate that in the final scene. That way it doesn’t get too overwhelming.
Sorry for the long answer, the more time I spend on forums the faster I write , it’s getting problematic 
1 Like
Hey !
I though it could be fun to test quickly what I talked about in the previous post :
You can look at the blend if needed :
the_witcher_test.blend (432.9 KB)
There are booleans, displace, remesh, shape keys on the logo.
For boolean and remesh because the settings where slowing down too much I duplicated them.
One is for the viewport with low res settings, the other is for rendering.
For the render, it’s a viewport preview , that uses the studio matcap shipped with blender.
About the result, I think it’s a good basis, but to get to a better result, more closer to the original, some more work is needed.
When working on an effect like this, I start to do some quick tests like here that generally take a few hours.
I could then work on the same element for another couple of days to get the most of the technique and add a polishing pass.
But in that particular cases, I’d probably try to have a rough version of everything first.
Like the archimboldo part, the finish etc…
Because even if that animation isn’t perfect, who knows , maybe with everything else it can do the job.
That way I get a first version of the whole animation in a short amount of time.
At this stage, I try to pay a lot attention to timings. The effects could be lacking , but if the timing is right, I can then refer to that latter. Because latter I may had other layers of complexity, changing the timings may be much more complicated.
Then I start to look at what particular element to work on first so the overall animation gets much better.
Once it’s done, I then work on the second most important thing to improve.
As said earlier the idea is to add several layers of meshs, effects, to a point it’s hard to tell what is what. Then the viewer is tricked : it looks more complex than it is and you don’t see individual elements anymore.
hopes that helps, good luck and feel free to ask questions !
1 Like