[Steve Hardie]

Might be worth mentioning that I had the same issue and was scratching my head. I did all of the above but realized thet the image was greayscale and not RGB, so even though it showed the chequered transparency etc it still was coming up as a solid in AE. Maybe obvious but thought it could help incase others come across this.

[Steve Hardie]

Thanks for your thoughts on this. There do seem to be a number of “workarounds” to this issue, although maybe they should not be classified as workarounds. My original post was working under the assumption that I was doing something wrong, when in fact there is no “right” way, but different ways of getting to a solution.

As I mentioned before, the method I adopted was to uncheck (disable) the active camera, which will then create an “ortho view”, I then manually reposition the image by matching the position and anchor point values ( this brings the image to center of the comp ( as otherwise it is randomly at the edge of frame) Once it’s in this position I can then “save frame as” and get the still. By then undoing the position change and reverting to the camera view, the 2d still that comes back in when checked as a 3D layer, aligns perfectly. Any other way seems to bring the 2D still back in as either too big or too small, which requires repositioning and resizing but seemingly by arbitrary measures whereas the other way seems to produce an accurate match.

The reason for this, as you note, is so that once trapcode has produced the lines I need, it is more efficient to replace the constant processing required by trapcode with a still so the continuing animation is more efficient. I also noticed that with the fine detail on these lines that the still image captured looks better (matches more closely) when saved as an RGB (without alpha) and brought in with a blending mode, as the prematted alpha version looks totally different.

Thanks for your thoughts.

[Steve Hardie]

Yes that is part of the issue; as I mentioned previously, I would have to turn the active camera off to get an accurate grab of the frame I need, but the frame (or the contents on it) are then all over the place which requires bringing the contents back into the frame by repositioning, then taking the still frame and then undoing those changes to revert all back to how things were. While I previously mentioned this I was not sure if that was “the” way to go or if I was missing something?

[Steve Hardie]

No. only using the one camera

[Steve Hardie]

Also the image I had sent was the wrong one. (attached correct here)
Could the expression work with a -1 so that the use of 0 would work? otherwise I will do as I said previously and only reduce scale to 1 and not zero and hope it is not noticeable.

[Steve Hardie]

Dan, “you are the man”, you must have a crystal ball over there. That is indeed the issue as to why it wouldn’t work.
I have to zero one of the nulls that rotae within a rotaion of another to make the particles scribe a circle and not a looping motion, but perhaps if I scale to 1 or a fraction, the expression will work, or at least be imperceptable?
Thanks, I shall have to visit your tip jar!

[Steve Hardie]

I think I wasn’t too clear. I am not turning the Particular layer into a 3d layer (it is already 3D in its own way). It is the resulting still frame that I am taking from the particular layer that I am turning into a 3D layer when that still frame is imported back into the comp so that it operates and matches the particular layer from which it was taken.

As noted above, the only reliable solution that I can do with some consistency is to temporarily disable the camera from the comp (which displaces the particular image off axis) and then to centre that by matching the co-ordinates, then save the frame. I then have to reposition the layer back to its original position, re-enebale the camera, and then when that still frame is brought back in and made 3D it will match the same position and scale of the original.

[Steve Hardie]

Thanks so much for your help.
If i make a new comp and apply this script it works great.
When I apply the script in my project it doesnt (see image) it returns an error, is there some other protocol I should follow when dealing with a project with multiple nulls?

[Steve Hardie]

does this work regardless of how many parent child relationships there are?
I have 3 nulls, the first two of which i want to scale but the third (last) in the chain needs the independent scale ( will move and reposition with the others but not scale) I tried the expression in the last child with this error “invalid numeric result (divide by zero?)”

[Steve Hardie]

It is an element in a sequence with a camera move.