Hi Friends,
I happened across the instructions for making a sunburst type of object, in Inkscape's Tips and Tricks tutorial, from the Help menu. So I decided to try it. These are the instructions, which I'm following precisely:

Choose the P1 symmetry (simple translation) and then compensate for that translation by going to the Shift tab and setting Per row/Shift Y and Per column/Shift X both to -100%. Now all clones will be stacked exactly on top of the original. All that remains to do is to go to the Rotation tab and set some rotation angle per column, then create the pattern with one row and multiple columns. For example, here's a pattern made out of a horizontal line, with 30 columns, each column rotated 6 degrees:

It should look like this:



But here's what I'm getting. Actually I'm trying for something about the same size as the above example, but I scaled my result quite a bit (~85%) to make it more suitable for inserting into message. But the rays are actually about the same size as the above.



The weird thing, is that if I use the exact same settings, but on a brand new canvas (doc), it works perfectly. What I can't figure out, is what's going on with the canvas on which I'm trying to use this object, that's causing it to be deformed as you can see.

Does anyone have any ideas what might be causing this? I realize you probably can't be sure without an svg file to examine. And I also realize I could simply make it on a new canvas/doc, copy and paste into current doc, which is probably how I will end up doing it. So this is certainly not a great, big problem

But it's really bugging me what might be wrong with the canvas/doc I'm working on Whatever it is doesn't seem to be affecting anything else in this image, it just bugs me to know it's there Plus, if I ask about it, I might learn something

Anyone enjoy solving mysteries?

I doubt it has anything to do with your canvas. More likely it's your object. If you create a new document and copy/paste that line from one to the other do you have the same problem?

Keeping in mind that I haven't seen the SVG, looking at your example gives me the impression your line is perhaps part of a much larger group where the other objects are invisible. The rotation will occur around the center of that group.

By "line" you mean the line that is the original tile that is to be cloned, right?

I tried copying the line (which is the original tile that I want to clone) from the canvas/doc in which I want to use the tiled clone, to a brand new doc/canvas. It cloned and tiled perfectly! Then I did it the other way around. I drew a brand new line, in a brand new doc/canvas, copied, and pasted into the doc/canvas where I want to use the tiled clone. Same result as what I showed above.

Keeping in mind that I haven't seen the SVG, looking at your example gives me the impression your line is perhaps part of a much larger group where the other objects are invisible. The rotation will occur around the center of that group.

No it is not part of a group. I drew it freehand, never grouped or combined it.

I think you once mentioned that one can use the Tab key to select one object at a time in an image? I think I'll try that, to look for invisible things.

Thanks for your comments, micro

A-HA!!
Well, as much as I understand what I'm doing, I've tried a few things to narrow down the source or discover the problem:
-- I made the background dark, to reveal any possible very pale objects -- none.
-- I used the Tab key to reveal every object, in case there might be an invisible one -- none.
-- I turned off the visibility of all the layers (3), then turned back on each layer by itself, one at a time -- nothing.
-- I tried to tile clones in different areas on the canvas -- no luck.
-- I tried to tile clones on eachh layer -- same deformed result every time.

THEN, I thought to create a new layer. Tile clones on the new layer, perfect results, just like on a new canvas/doc! So now we know that there's something about the 3 layers already in existance when I 1st tried the radial tiling.

With this new info, can anyone deduce or even surmise, what's causing the deformed tiling? I might mention that I've played around with simpler tiled clones, and have found it quite frustrating. Most likely, I just don't understand it well enough to use it successfully. But I have to wonder if the problem I've run into in this image, could have fouled up my previous attempts (at much simpler tilings). The problem does seem particular to the 1st 3 layers of this image alone, though....

Anyway, any thoughts, ideas or theories?
Thanks for your interest

One thing to check as microUgly mentioned is to check where the rotation point is (I guess that would copy over to a new doc so it's probably not that).

In the Create tiled clones dialog do you have Use saved size and position of the tile ticked? I think it might be safer to have it on.

I'm not exactly clear about this issue, I experienced similar problems, I had tiled clones in a circle and when I copied them to another doc they flew apart all over the doc (still forming a circle but a huge one). There is one option in the Inkscape preferences > Clones called "When the original moves, its clones and linked offsets:" and you can choose Move parallel, Stay unmoved and Move according to transform (this is in dev version). I think it might be related to the Move according to transform option. If your original object has some transformations done on it, it may be that its clones pick those up (inherit them) so you have additional functions performed on the clones in addition to the tile clones command.

Maybe the Use saved size and position of the tile in the tile clones dialog removes the transformation info from the object before tiling, I'll test all this when I have some free time, or maybe a developer can clear this up a bit.

In the Create tiled clones dialog do you have Use saved size and position of the tile ticked? I think it might be safer to have it on.

No, it's not ticked. I'm not entirely clear what it does, to be honest
Ok, I've just tried again, moving the original tile/line to each layer for each trial; this time with the 'Use saved size and position' checked. But again, it only works correctly on the new 4th layer. Same deformed result on original 3 layers.

If your original object has some transformations done on it, it may be that its clones pick those up (inherit them) so you have additional functions performed on the clones in addition to the tile clones command.

The original tile/line has not been transformed. I've created a brand new line segment with each new trial -- many by now

Is it possible the 3 "bad" layers carry some transformation info that's affecting the tiling? Because in this image, I did a transformation, (skew) on one object in one layer (NOT the original tile/object/line). But I don't know how it would affect a brand new object, each new original tile/line I've made for these many trials. Ironically, I'm planning to skew the successfully radially tiled clones

There is one option in the Inkscape preferences > Clones called "When the original moves, its clones and linked offsets:" and you can choose Move parallel, Stay unmoved and Move according to transform (this is in dev version).

Yes, I have the same settings in my 0.46 version. It's set on "Stay unmoved".

Thanks for your coments, prkos

[Edit] As an afterthought, I tried moving the correctly tiled clones on the new 4th layer, to one of the 3 "bad" layers. And it's just like you said happened to you. When I simply moved it to another layer, all the clones came apart, and were arranged in the same deformed way, as if I had tried to tile them on that layer to begin with. So unless I'm ok with having the 4th layer in my completed image, it's not going to work as I had hoped, i.e. - copying the correctly tiled clones from a new doc/canvas or new layer, and pasting into my image -- unless I want a new layer....

It might be best if you post the file online for examination... perferably not on PetaIMG if you want a quick response from myself

If your original object has some transformations done on it, it may be that its clones pick those up (inherit them) so you have additional functions performed on the clones in addition to the tile clones command.

If you in preferences have checked Transform - Store Transform - Preserved (instead of Optimized),
you will get the behavior described, if the original object has been moved.

If you in preferences have checked Transform - Store Transform - Preserved (instead of Optimized),
you will get the behavior described, if the original object has been moved.

Oops, I think I said some nonsense here. Please ignore it.

Aaarrggh!
Sorry micro, I just don't have another alternative to petaIMG right now.

Ok, what's left of my trials with the radial tiled clones are to the left of the image. The image is inside the page border, and the 2 tiled clones are to the left. The tiled clone that looks correct is in Layer 4. And the deformed one I believe is in Layer 1. And just to be clear, Layers 1, 2 and 3 are in question -- if the original tile/line is in one of them, the result is deformed. Here's the link:

http://www.petaimg.com/uploads/1223783430.svg

(FYI -- It's sometimes hard to open images hosted at petaIMG. Often you'll get an error message saying the image must have been removed, or the link is bad, or something else I don't remember at the moment. Just keep trying, eventually you'll get it! And my apologies, I just don't have another means of hosting .svg files right now )

And thanks again, as always, I appreciate your comments

Hi Brynn.

Nice beach creatures in the file

In the XML Editor is a clue to the conundrum: All the "bad" layers have a "transform : translate[..]" attribute, where the values for translation are not near zero (the fourth layer has a translation attribute that is 10^-6 so negligibly small. Layers created in a new document dont have the transform attribute at all, i am not sure whence it came from...

If you delete the transform attribute, tiling works as advertised -

btw: i see no worksheet in your data - how did you do that?

All the "bad" layers have a "transform : translate[..]" attribute,
...

That was what I intended to say above. But a while after running a test, I noticed that it was done with P6 symmetry instead of P1. So I was just feeling too stupid for a moment to test more.

Some time ago, I tried to fine adjust the tile parameters by editing them in XML. This was done in order make a "perfect" hexagonal tiling of a bitmap. Doing that, I noticed the importance of the transform:translate attribute, but I can not remember the details.

One more thing: There is a exclude tile checkbox on the shift tab. Have you tried setting that instead of shift -100%?

Thanks for your help loonquawl and Naga.
I had used a skew transform on one object in one layer. But I don't understand why the transform data has been attributed to the layers. Is that how it's supposed to work, when you transform an object, the layers carry that info? Wouldn't it be more appropriate for the object to carry that data, rather than layers? Please note that I only have the vaguest of understanding of xml, in case that sounds like a stupid question

btw: i see no worksheet in your data - how did you do that?

I'm afraid I don't know what you mean -- what's a worksheet? By data, do you mean the info in the xml editor?

One more thing: There is a exclude tile checkbox on the shift tab. Have you tried setting that instead of shift -100%?

No I haven't tried that. In this case, I was trying to following the instructions precisely, and they said to use the -100%. But that's good to know about, now that you've pointed it out

Thanks again for your help guys

btw: i see no worksheet in your data - how did you do that?


I'm afraid I don't know what you mean -- what's a worksheet? By data, do you mean the info in the xml editor?

The virtual "sheet of paper" that marks the working area in Inkscape. It usually is A4 oder Letter or whatever Default you use - i couldn't see that in you file. Is it possible you imported an svg from another application of somesuch?

Oh, could that maybe be because after I finished, I selected the entire image, went to Doc Properties, and clicked "Fit page to selection"?

Interesting feature. What do you use it for? It introduces the transform attribute to the layer - any layer added afterwards does not have the attribute - no idea if this is bug ot feature

Whoa!
Are you saying that using the "Fit page to selection" feature causes any transformation, anywhere in the image, to assign the transformation data to the layers? Even though the transformation data does not affect anything except the object(s) to which it was applied, it appears in all the layers?

The reason I use that feature is to have my page size exactly fit my image. Otherwise, my finished product is a big white rectangle with my image only a small portion thereof. And if I upload it without making sure it fits, all or part of my image might be cut off by the page border. Basically I do it to make sure all my image is within the page border. Because it annoys me to work with the border visible (and I especially dislike the shadow).

But if we have stumbled upon a bug, maybe I shouldn't use it?

Thanks for working on this, loonquawl

FYI for anyone who might be following this or search it out -- the info about the transformation data being assigned to the layers, and causing the distorted radially tiled clones, has been added to Bug Report 284709 as a comment. Still yet to be "confirmed" or "assigned", at this time.