That back bending is simply unrealistic. Just changing the bending 
direction already improves things. What I did was add 180 degrees to all 
negative rotations.

Thanks for all the suggestions.

@ Bald Eagle: I do indeed use the image_map as a kind of force map, 
through the height_field stage and then using the properties of the 
normals. Basically, I think that is identical to what you suggest? The 
rate of bending is proportional to the normal: pointing upward => no 
bending; increased bending with increasing normal angle towards the 
horizontal.

@ MichaelJF: Yes, that parting was killing the effect and came from my 
concept of rebounding of the stalks. In any case, this was far to strong 
here. I presently dropped this effect for the time being.

@ clipka: the bending angle is uniform over the whole length as a first 
approximation. In reality, I expect bending to increase towards the top, 
due to the different effects already mentioned by Alain and Bald Eagle. 
Most important in my view are the thickness/rigidity of the stalks 
towards the base and the damping effect of surrounding stalks. The 
former is easy to implement, the latter is more difficult, especially if 
one also wants to consider collision effects. At present, I do not know 
how to tread that path.

Something I still believe not right is the wave length of the effect. I 
think it should be larger.

Also note in the present image that all stalks start from the same 
initial orientation/scale. I need to introduce more randomness /and/ add 
different types of stalks.

Thomas

Post a reply to this message

Attachments:
Download 'hf_pattern_testing.png' (576 KB)

Preview of image 'hf_pattern_testing.png'