I will need, in one scene in my piece, a t shirt with quite a bit of text on it. From my previous venture into using the "text tool" in ASP, I foresee quite a nightmare trying to do the shirt that way. So I had this idea, see. What if I create the text in photoshop, and import it into ASP, and manipulate it there (perspective, scale, etc.--there will be three angles on this character, one from the front and a three quarter one way, and a three quarter facing the other way).
This seems like it would keep out all the horrid glitches that the text created in ASP presented me with, and as long as I make the photoshop text the right size, shouldn't pixellate or otherwise look hideous when brought into ASP.
So, if I'm going to have an HD project (1920x1080), and I just make the text in photoshop HUGE, if I import and start scaling it as appropriate in ASP, will that work?
Or should I figure out the exact size(es) I need based on the actual shots in the movie, and then create the text in photoshop to that specification?
Thoughts?
text on clothing with image layer?
Moderators: Víctor Paredes, Belgarath, slowtiger
That trick should work fine. As for the size don't make it too big. However if you zoom in on the character or shirt you need to make sure you have enough resolution so it holds up.
If you desperately need to save on render times I would make the image is as big as it shows in the animation... plus a little extra. If you plan to zoom for close ups you can always make a larger version. For small text you would probably want to use a larger image to keep it readable.
You should probably do some tests. Create the photoshop text image and experiment at different sizes in AS and render them. Check to see if the quality meets your standards... then drop it down till it doesn't... find that sweet spot between quality and size.
Of course the photoshop document will have "live" editable text so you can scale it up and down as much as you want without worries.
You could put the image in as a texture in a shape or image layer with bone warping to make it conform to the shape of the character.
------
I am currently working on an AS project that is ALL images and text. Animated books. The whole thing is just image layers. The images are 22" x 17" at 72 dpi. The rendered video is NTSC. It looks fantastic. To be honest the images are probably bigger than they need to be but the render times aren't too bad... and... well I don't have to render the files and the images were provided by someone else.
-vern
If you desperately need to save on render times I would make the image is as big as it shows in the animation... plus a little extra. If you plan to zoom for close ups you can always make a larger version. For small text you would probably want to use a larger image to keep it readable.
You should probably do some tests. Create the photoshop text image and experiment at different sizes in AS and render them. Check to see if the quality meets your standards... then drop it down till it doesn't... find that sweet spot between quality and size.
Of course the photoshop document will have "live" editable text so you can scale it up and down as much as you want without worries.
You could put the image in as a texture in a shape or image layer with bone warping to make it conform to the shape of the character.
------
I am currently working on an AS project that is ALL images and text. Animated books. The whole thing is just image layers. The images are 22" x 17" at 72 dpi. The rendered video is NTSC. It looks fantastic. To be honest the images are probably bigger than they need to be but the render times aren't too bad... and... well I don't have to render the files and the images were provided by someone else.
-vern
You need to take extra care in creating this text in Photoshop, especially with spacing and anti-aliasing settings. If you distort the text in PS, be sure to use "perspective" distortion, otherwise you'll have seriously incorrect proportions. Be sure to only use a licensed font with all necessary characters, especially when you produce all the different text versions in foreign languages for international distribution. For maximum realism, restrict yourself to silk-screen-save colours for the text on the T-shirt. It helps when you pre-shrink your character at 60° C before you start animation.
*coughs*
Seriously, how important is that writing on the shirt? Does it contribute to the story? How much time do you spend in making that little item look right, compared to overall animation?
*coughs*
Seriously, how important is that writing on the shirt? Does it contribute to the story? How much time do you spend in making that little item look right, compared to overall animation?
Looks like that's geared toward more textured clothing. . .
My characters are about as flat as can be:
http://www.youtube.com/watch?v=jb6iJQB666E
My characters are about as flat as can be:
http://www.youtube.com/watch?v=jb6iJQB666E
After seeing your sample (I really like it by the way!) you may just want to use vector text in AS. How much text is there? If it's a LOT of text you won't be able to read it anyway. If it is "t shirt" text with a few words then doing it in AS would be the way to go in my opinion.
I did this tshirt design in AS. I distorted the points to match the... uh... contours of the character
(click for larger image)
-vern
I did this tshirt design in AS. I distorted the points to match the... uh... contours of the character
(click for larger image)
-vern
I am flattered that Vern likes my stuff. That's like Zakk Wilde telling you you're an awesome guitar player.
Anyway, yes, there will be a lot of text (15-20 words or so). There will be a close up tilting/panning shot that sort of follows the text as another character reads it, not really intending it to be readable in a wide shot, but maybe a medium, depending on the font size. That CU will definitely happen, so if it's not readable in any other shot, that's okay. But for continuity's sake I'd like it there regardless.
I had a hat on another character with some writing created in ASP on it (not shown in my demo, 'cause it's my super secret awesome villain), and that was a pain to deal with. . .I kept getting glitches in the letters (filled in holes in Ps, Rs, etc.) that at first I thought I'd fixed by using a different font, but found I was wrong. Just seems like for this the image might be the way to go.
Anyway, yes, there will be a lot of text (15-20 words or so). There will be a close up tilting/panning shot that sort of follows the text as another character reads it, not really intending it to be readable in a wide shot, but maybe a medium, depending on the font size. That CU will definitely happen, so if it's not readable in any other shot, that's okay. But for continuity's sake I'd like it there regardless.
I had a hat on another character with some writing created in ASP on it (not shown in my demo, 'cause it's my super secret awesome villain), and that was a pain to deal with. . .I kept getting glitches in the letters (filled in holes in Ps, Rs, etc.) that at first I thought I'd fixed by using a different font, but found I was wrong. Just seems like for this the image might be the way to go.
Okay here's my suggestion...
Use an image for the text, but treat the close up as a separate shot. Don't think of the text as a single thing to be used the same way. If you only have one close up shot then treat it as single specific shot and use a higher resolution image.
For the close up image version you may just want to use the tools in Photoshop to distort the text. Maybe not have the character wearing the shirt move so much while the other one is reading it. This can save a lot of effort in AS trying to distort the shirt text.
In the rest of the animation just use a smaller more manageable image layer. The same text just a smaller size. It isn't going to need to be very large. Also considering the simplicity of the character designs simple distortions with a few bones would work. I don't think distorting the layer will work so well. You are limited to what you can do with an image layer. You can only rotate or scale it. You can't bend it or anything like that without bones.
With one or two bones over the image layer you could do enough distortion to get what you need I think.
-vern
Use an image for the text, but treat the close up as a separate shot. Don't think of the text as a single thing to be used the same way. If you only have one close up shot then treat it as single specific shot and use a higher resolution image.
For the close up image version you may just want to use the tools in Photoshop to distort the text. Maybe not have the character wearing the shirt move so much while the other one is reading it. This can save a lot of effort in AS trying to distort the shirt text.
In the rest of the animation just use a smaller more manageable image layer. The same text just a smaller size. It isn't going to need to be very large. Also considering the simplicity of the character designs simple distortions with a few bones would work. I don't think distorting the layer will work so well. You are limited to what you can do with an image layer. You can only rotate or scale it. You can't bend it or anything like that without bones.
With one or two bones over the image layer you could do enough distortion to get what you need I think.
-vern
Makes sense. In my mind's eye, the guy with the shirt won't be moving at all while the read is going on. . .he's just standing there.
As for the deforming of text, are you saying the shear, perspective, curve points H/V, those kinds of tools, don't work on the image layer? I hadn't thought of that.
I may try the bones thing, but if my guy with the shirt doesn't bend his torso (he only wears it for a short portion of the movie, and then uh. . "transforms" into an entirely different entity where the shirt is a non-issue), maybe I bring a still of him into photoshop and match the text perspective in THAT program, bring it into ASP as an already distorted image that I never intend to deform later.
As for the deforming of text, are you saying the shear, perspective, curve points H/V, those kinds of tools, don't work on the image layer? I hadn't thought of that.
I may try the bones thing, but if my guy with the shirt doesn't bend his torso (he only wears it for a short portion of the movie, and then uh. . "transforms" into an entirely different entity where the shirt is a non-issue), maybe I bring a still of him into photoshop and match the text perspective in THAT program, bring it into ASP as an already distorted image that I never intend to deform later.
Sounds like a plan. I tried putting a text image layer over one of my characters. The problem is the bone settings. I use region binding which is terrible with images. They bend "hard" instead of "smooth". To bend image layers you really need to use flexible binding... which would require just a couple of extra bones and using more bone offset with the character I was experimenting with.
I could make it work great if I just offset the image layer (using bone offset) so the bones from the rest of the character wouldn't effect the image so much.
I've done this before. Slide the image away from the main body skeleton. Add a few bones to control just the image, then using bone offset slide those bones back over the body.
What I do is to use bones that sort of match the spine or back bones of the character skeleton then add rotation constraints to the image bones that target the skeleton bones. Now the offset image bones bend just like the body.
EDIT: You don't have to use constraints if the bones over the image are children of the bones in the character skeleton. After offsetting they will move just like the parents.
I am simplifying my explanation but that's the general idea.
I used a fairly average size image from photoshop saved as png with alpha and zoomed in real close. Couldn't even tell it was an image at all. Blended right in with the vecor art even when rendered. You shouldn't have a problem with this technique.
-vern
I could make it work great if I just offset the image layer (using bone offset) so the bones from the rest of the character wouldn't effect the image so much.
I've done this before. Slide the image away from the main body skeleton. Add a few bones to control just the image, then using bone offset slide those bones back over the body.
What I do is to use bones that sort of match the spine or back bones of the character skeleton then add rotation constraints to the image bones that target the skeleton bones. Now the offset image bones bend just like the body.
EDIT: You don't have to use constraints if the bones over the image are children of the bones in the character skeleton. After offsetting they will move just like the parents.
I am simplifying my explanation but that's the general idea.
I used a fairly average size image from photoshop saved as png with alpha and zoomed in real close. Couldn't even tell it was an image at all. Blended right in with the vecor art even when rendered. You shouldn't have a problem with this technique.
-vern