Foreshortening

Post a pic of what you're trying to achieve.

Try this on for size...

Shorten.ffxml

I think the 3D term for what I am trying to do is called the 'perspective plane'. Right now I am working on the x axis but I want to add y and z also. Z should be trivial and y should be much like x but mathed together some way.

Just imagine placing a soup can at any location in a photo while matching the perspective of the photo.

I want to move a want to allow the user to place a cylinder in a photo.

Well a cylinder would be interesting. You might want to take a look at some of these filters to figure out shapes...Some 2d and 3d methods...Maybe take some of these and rework them to something new or just for the study...Or they might just give you some direction or a new idea..

http://www.filterforge.com/filters/2636.html

http://www.filterforge.com/filters/1188.html

http://www.filterforge.com/filters/1849.html

http://www.filterforge.com/filters/587.html

http://www.filterforge.com/forum/read...10&TID=747

sounds like a skew/depth perspective, isometric view thing. i would think your background then, would be transparent. you import your soup can image and then 'tilt' it forward or back with your FF controls to give the proper perscective to the can. have i got that right?

Another use is for 3D artists -- where they apply FF images to 3D objects and their 3D program creates the perspective for them. (I think this is where the feature of seamless tiling becomes really handy -- i.e. when the image is used as a texture.)
By the way, these 3D programs can also be used for adding in objects to 2D photos. In case you're interested: Blender is a free, versatile 3D program -- but it does seem to have a steep learning curve at first.


It would be cool if FF could do perspective skews -- maybe it's somehow possible (but it seems easier to use other programs for now).

ronviers,

Quote
Is that the way it works? Do most of the FF users also users of 3D authoring software? What would your estimation be of the demographic breakdown for FF users in terms of bitmap, 2D vector and 3D? Who are we making these filters for?

Actually, I have no idea. The person who mentioned FF to me was a 3D artist -- and FF seems like it could be very helpful for those types. (So this is just one way it can be used).
Meanwhile it does seem like a lot/most of the people here at the forum deal more with just 2D bitmap stuff -- where it could take some time before more 3D artists discover FF (…?).


As far as 3D programs go, I don’t have much experience beyond experimenting with free trial downloads. One trial program not on your list that I liked was Zbrush. It’s not really a "full" 3D application, but it seems to have some tools that make modeling (or whatever) fun. By now, I think there are similar copy-cat programs that work similarly too (…which seems to suggest that their ideas are good).

From my experience, most normal 3D programs work pretty similarly -- so if you learn one, it’s not too hard to learn another. I’m mentioning this because it may be easier to try out a free trial version of some non-free (expensive) software (that has a good help manual) before trying to understand Blender -- but maybe not.

It’s been a while since I’ve explored 3D -- but I think at some point I may return, as it seems to have a lot of potential…

ronviers,

Quote
I was speaking with StevieJ recently about how stylistic differences in filter creation were a mirror of our personalities but now I am beginning to think it may who we imagine our respective target audiences to be that mostly determine the outcome of our filter designs. Approaching FF from the POV of graphics design, photography or gaming/cinematography is bound to play a major role.

Interesting.
Filter creation is an art form IMO. I just create filters for myself -- and then submit the ones that I think there’s a (slight) possibility that others might also find it useful. However a “great artist” probably is more integrated, where they can create filters from their heart, while also creating something that others would most likely also like.
I guess it’s a balanced mixture -- personality, where you’re coming from/your POV, and what you think others want… I admire the filter creators that seem to easily create "great" filters.

Quote
That sounds like excellent advice. Lightwave3D is the only one I checked on and it has a 30 day trial but I have not given up on earning a copy of FF so at this point I do not want to start a new trial.

Yeah, since there are time limits on most of the trials, it’s probably best to just focus on one program at a time. (I was lucky where I played with FF in the beta phase -- where they were very generous in giving out free programs.)

Quote
I either need to nail this placement issue or move on because I have spent too much time already trying to figure it out. I actually dug out a trig book and a linear algebra book last night. Something tells me I’m getting way off track if I am working problems in text books. Focus is a huge issue for me.

That’s cool about looking up math books. It sounds like something I would do (and have done for other things in the past).
I was thinking about how to create a nice, perspective skew in FF -- but then decided it was too complex. If you pull it off, it would be uber-cool (innuendo intended), but as far as trying to earn a free program, there might be more efficient ways to go about it. (?) (But I still say, follow your heart.)

Quote
This seems like a good time to get started in 3D- and if you do I hope you will let me know which direction you take.

Yep, 3D programs are really powerful, and these days computers are getting tons of ram and speed… Right now, I’m moving more in a 2D -> animation/film direction, but it’s possible that 3D will become more appealing.
If I ever come up with any thing cool, I’ll probably post it in the user gallery -- but to be honest, I’m not that great of an artist -- where I like to mostly just fiddle as an hobbyist and admire what others make.

Interesting – very very interesting. A few details remain but progress at last!
Excellent work!

P.S. I just thought that the filter might be a little harder than I first thought to understand, so I'll add a brief explanation.

I used two zoomer snippets (developed by Lucato and modified by uberzev) -- where one is supposed to shrink an image and the other enlarge the image. Then I used a gradient mask so (as I originally pictured it) the top of the image shrinks, while the bottom expands (giving a perspective skew of sorts).
One problem is that when the image shrinks, it causes multiple images to appear (so maybe a mask of the non-center parts is needed).
Another problem is that the shrinking doesn't go very extreme in this filter.

oops!

Here it is

Scale working.ffxml

Now with scale up!

I still have down issues with masking and rotation but I do not think they are serious.



Scale temp ii.ffxml

I like how small you can scale down with your technique.


Here's my lastest perspective attempt where you can rotate (thanks to uberzev's Rotate Any Angle snippet) and skew the plane. (I'm mainly focused on the center image, where I manually mask the borders to remove the distractions.)
It's still not perfect, but kind of interesting (to me at least).

perspectiveskew5.ffxml

Thanks, Ron.

I like the wire frame mode idea -- but I'm not sure what can be processed faster than the image itself. (Actually maybe the checkers are faster? But I don't know...)

The two main improvements I think my filter needs is that: 1. it can scale down smaller and 2. it doesn't warp when the small and large proportions are extreme. I think the trick will be to come up with a better scaling method -- where I may have to analyze your way more. (But I may end up getting too side-tracked from this project to finish it...)

Quote
I had some preconceptions about how I wanted skew to work but they turned out to be wrong. So it’s one step back for me - I can be so thick headed.

Ha. I can identify a lot with that.
But I think experience slowly starts to help in this area. Most of my ideas for that last filter occurred to me while daydreaming at work -- where I was surprised how well they turned out when I finally got to try them in FF. Normally for this kind of stuff I have to do a lot of trial and error (especially when predicting how gradients effect offsets).

I finally got the scale to work perfectly across the entire range from all the way small to all the way large. I also got the masking to work at all sizes and in all positions, plus I added a knockout background. Even without the skew it is a pretty powerful filter. It will be great for placing my glass hearts.

Thanks

Scale_Shift.ffxml

Magnification, for me personally, has connotations of spherical aberration so how about this one:

Resize_Move.ffxml

Cool. Nice job getting an auto-mask to go with the shrink effect, Ron.

Look at how screwed up my rotate is. It is the top half of the components. Two and a half days and this it all I have. The way I envision this working is something like a helix on end that is pushed through the offset component with the slider. The idea is that the four impulse components make up a central pivot point while the circular arc components provide a correction from center to edge. The problem is that none of it works.
Before I started this I searched the FF filter library and found a ton of rotate filters so I assumed it must be easy to do – I mean, if everyone is doing it. But now I don’t think it is all that easy. I’m not giving up, just frustrated it is taking me so long.


w_Rotate.ffxml

It has been one of the most enjoyable problems I have ever worked on.

Quote
doing it that involved sine and cosine math

Yeah, it has come to that. I was just reading in my trig book about converting from rectangular to polar coordinates. It looks like square root may be a problem. I wrote a square root program for my calculator years ago but it was iterative so I don’t think it will apply to FF.
I hope to get a large chunk of time this weekend to work on it but real life is always butting in.

This top group (y squared) shows what I *expected* to get from the previously mentioned configuration. I’m not sure what’s going on here but I was noticing the graph of the square root of x appears to be identical to the graph of x squared but at a right angle. So it may be that square roots can be derived simply by placing curves and gradients at right angles to one another.

This weekend turned out to be a bust so I will get to look at it again some time next week.



Polar.ffxml