Hello
I am trying to get my head round how to setup a scene that will deliver an anamorphosis effect such as the giant “3D” billboards that have been popping over the years.
There are plenty of examples on youtube but the link below shows the anamorphosis effect from3 angles
Any thoughts are most welcome.
Thank you
The screen (especiallyt rounded one) delivers an image which is best seen from a specific angle (the camera in the scene)… so projection modifier (from the scene camera) of a (flat) animaton on an possible rounded object (whilst the animation is “filmed” from a similar scene with similar camera looking in a similar way)…
see also:
Thanks for the reply.
I suppose that my question is do to with the 3D scene itself.
The “forced perspective” effect deforms the proportions of known objects so that from a certain angle the illogical deformations look coherent. In the Cat example, I presume that the paw was disproportionately large in comparison to body, to exaggerate the optical parallax effect. Therefore I presume that the 3D model of the cat was deformed whilst keeping the animations very real.
It strikes me as very delicate deformation technique but I am probably very wrong.
Also, the effect probably needs to be seen at some sort of scale so I wonder how to simulate the effect on a desktop monitor.
One trick is that the screen is “not used by its full size”, the box gives an inner frame and breaks out, so you see the surrounding parts of the screen as part of the building. Also so you can put parts of the the cat out side the box and give just more an 3d illusion effect.
I don’t know if having a 2 sided screen is a prerequisite for the effect, but framing the action in a box definitely adds loads of perspective to the effect . The resident evil example is great for this
Here is a side view of the panel and the amorphosis stretching effect is not perceptible like with the cat.
I presume that as the animation is very dynamic, using all the space, the brain compensates to give a 3D feel.
Is the 3D “wrapped” onto the screen in 2D to give the false perspecive?
It looks like any high-resolution curved billboard with LED’s can do this (because of the viewing angles possible along with its headroom for distorted imagery). The back view shows there is holographic 3D display at work here.
I am not sure what you mean by these popping up, because I do not think they are too common in the US yet (my hometown also has less than half a million people, so we do not have a large skyscraper to mount the thing to).
I would
OP means Japan, probably. I am certainly not going to see one of those either unless I travel very far
Hadriscus,
I’m not sure that it is that simple. If you look at the resident evil examples, there is loads of foreshortening that can be seen better on the side view. It is the foreshortening effect that gets me. Distorting the animation to give the “out of the screen” effect.
This reminds me of these “portal based shaders” where you can recreate a cubical room on a flat plane.
Which is a 3D space on top a 2D surface, but with pumped projection in order to appear as if is a real 3D space in the eyes of the viewer.
Can you go somewhere with this or is it entirely different?
It’s a bit different because the shader is dynamic, what is “drawn” on the plane change according to camera view. Where that’s something you can’t do with these billboards.
The billboards are very similar to these anamorphosis street art thing :
There is an ideal point of view and from there you fake perspective and create some depth illusions.
And the more you go away from this ideal POV, the more distortion you get :
The shader works a bit differently, and it allow the ideal POV to change depending on the camera angle, so you always see a correct result.
While both are optical tricks, one doesn’t help much to make the other.
Thank you for your thoughts.
Creating a "deformed " or forced perspective in a painting is a challenge , it seems way more difficult to animate while keeping a visual coherence. The exams have fast animations that probably cover up the deformation but it still seems a real challenge. And how can you have an idea of the result on a desktop monitor. I figure yesterday scale is very important.
That is where @Hadriscus step 4 comes in.
By making your scene in full 3d, without trying to distort anything, you’ll have a much easier time.
Then you can project that image onto your screen, the projection does the work of the distortion for you.
I think that’s just baked into the animation. I don’t believe there is any ‘holographic’ display tech going on here at all.
This means that you will need to replace the 3D viewport matrix with another object matrix, typically you set the camera (or any other fixed object).
If you see this image from drone, the display completely awful. But from street view everything looks as intented.
Here’s a quick mockup to demonstrate the concept:
3d scene:
take care to keep all of your objects within the bounds of your display, as seen from the camera
here is the rendered scene:
now, using a uv project modifier on a mesh representing the screen, we can project the image onto to screen:
note that there is some UV distortion on large planes, so subdivide the screen mesh a few times, as needed:
as noted, the effect really works in one spot. change the angle and it’s broken:
Then, setup a second uv map for the flattened version and bake the texture:
Hope this helps @chlowden007
Great, do you think this will be feasible to be done in one-go? In the new viewport compositor.
I have thought of some render-to-texture workflows in order to get 100% more optimizations in rendering. Going through the render-and-rerender workflow is very cumbersome now.
Maybe? I haven’t looked at the new compositor, though. I’m not sure about the baking to texture part, in particular.
We will have to see. As for example now the most easy answer is to run two instances of blender and then reload the texture at the other instance. It looks like over the top solution…
Yeah, that seems like doing a lot of work to avoid a little work, haha
