CS5 - Wrap label around a cylinder (like wine label or pop can)

[tomaugerdotcom]

[This is not a Windows-specific question, I hate to exclude Mac users from this answer, but there is no platform-neutral forum here]

In Photoshop CS3 and earlier, taking a layer, such as a label for a wine bottle or jar, wrapping it around a 3D cylinder and compositing it to an image (of the actual jar or bottle) was almost trivially easy.

Unfortunately, with the advanced 3D tools available in CS4+, the simplicity of the Wrap > Cylinder option seems to have been lost.

Wrapping a complex texture around a cylinder (full coverage, top to bottom) is possible, but how about wrapping a label (especially one that has a die-cut ie: not full coverage)?

A couple of quick searches have surprisingly turned up no tutorials on the subject. Has what was once such a simple technique become so technical and obscure?

Can someone point me to a tutorial (written or video), or provide a list of steps here on this forum?

Thanks!

[tomaugerdotcom]

I should probably point out that I'm not looking for a hack or simulation like using Free Warp, or Shear on the visible half of the label. I would like to see how to do it using the 3D tools:

1. adjusting the scaling of the label, and avoiding distortion

2. changing the positioning of the label relative to the reference image (ie: how high or low it sits on the jar or bottle)

3. in the case of irregularly-shaped labels (ie: with die-cuts) how to ensure transparency that matches the shape of the label

4. how to match the 3D primitive / camera angle / perspective distortion to the reference object in the photograph

All this used to be so easy and now seems like you need a background in 3D rendering and animation in order to do the same tasks.

[Noel Carboni]

I don't have Extended, but there was a thread here a while back where labeling a wine bottle was discussed I think....

Ah, here:  http://forums.adobe.com/thread/791137

-Noel

[tomaugerdotcom]

Thanks Noel, that is a fairly detailed post. But it doesn't specifically address irregularly-shaped labels (ie: die-cuts)...

[SG...]

Are you looking to do something along these lines? The glass is a photo and the text label is a 3D object.

[SG... · Accepted solution]

Hi,

Outline for this exercise is to generate a 3D cylinder, create artwork you want to wrap around the cylinder, duplicate and modify that artwork to be used as a mask, and then arrange the 3D element on an image to composite together.

So the basic steps I used to do this:

1) New 1024px x 512px, RGB doc w/White background.

2) 3D> New Mesh from Grayscale> Cylinder.

3) Window> 3D. Select the material 'Background'.

4) In the lower section of that 3D panel, select the Opacity texture pop-up and choose 'Remove Texture' (you will be adding one back in step 10).

5) In the same area select the Diffuse texture pop-up and choose 'OpenTexture...'

6) Here's where you add the label graphics that you want bending around the cylinder. I used a text layer w/stroke layer style for this example.

7) When you have the artwork as you want it, Layer> Duplicate Layer... to a new file.  This new file will be used for the Opacity texture which masks off the area of the label to give you irregular shapes (ie. die-cut).

8) Select all the layer pixels and change them to white, then on a new layer below fill it black. This is your new Opacity map. Save it to a working directory for use in the next step, and close the file. Also save and close the artwork file from step 5/6 (it is stored as a smart object in the original file from step 1)

10) back with the original file from step 1 (with the 3D layer selected) in the lower section of the 3D panel, select the Opacity texture pop-up and choose 'Load Texture...'. Browse to the opacity texture from step 8 and choose 'Open'.

11) You should now see just your label artwork wrapping in a cylindrical shape. You can use the 3D Camera Rotate tool to view the various angles.

12) now, bring in your jar image for compositing. This is where manipulating the 3D layer to align the scale and perspective can get tricky. I try to just use the 3D camera tools for this work (with global object scaling an exception). I also use the 'Shaded Illustration' Rendering preset to help visualize.

Hopefully this gets at some of your questions. Here's a related Dr. Brown title for CS4 that demos the interaction and re-enforces the texture mapping aspects: http://tv.adobe.com/watch/the-russell-brown-show/creating-transparent-3d-globe-graphics/

regards,

steve

[tomaugerdotcom]

Right on Steve, that's exactly what I was looking for. If you could just answer two more questions I think I'll be well on my way!

1) when I've tried this technique in the past, the artwork has been scaled / stretched or otherwise distorted disproportionally as it wraps around the cylinder. What steps do you need to take to ensure that the artwork not only maps on, but scales appropriately.

2) Further to that, assuming you can get the proportions right, what do you need to do to change the scale / position of the "label" on the cylinder? The u/v controls are a little kludgy - I was hoping for a drag and drop texture repositioning tool of some sort. Also, when I reduce the scale of something (by increasing the uv scale values, the texture then tiles repeatedly, and I couldn't seem to find any options to turn repetitions off. So what would the process be if you had, say, a really large vector art file for the label and needed to scale it down so it only occupies a portion of the cylinder surface?

Again, thanks for the detailed walk-through!

Tom

[SG...]

Hi Tom,

I'm glad that was on target on didn't create more confusion ;-). I was a bit nervous without visuals to help illustrate my written description.

1) That is a very good question and a frustrating problem when you run in to it. The issue comes from how the geometries are created, their UV mapping, and the texture map's proportions. Looking at my samples above, I see there's distortion.

For this case using the New Mesh From Grayscale> Cylinder command, you can do a couple things:

• The quickest would be to use the 3D object scale and modify the Z value to control the distortion. Since you're hiding the bulk of the cylinder mesh, and using a right rectangular cylinder scaling the Z value works well to make the mesh and UV mapping square.

• You can also open the texture maps and transform scale them to correct for the non-square mesh. This involves at least 2 maps (Diffuse & Opacity) so It's not quite as clean, and it's a bit more iterative since you need to toggle between docss to see the changes.

• You can also experiment with changing the image size of the texture maps, so when they map to the cylinder mesh the resulting display is not distorted.

• The command to generate the cylinder uses the color values in the starting flat file to define the radii, and this also affects the slope of the resulting polygons. For 8bit RGB documents, using White (255, 255, 255) results in mesh triangles with a slope of ~.5 (27º), mid Gray (128, 128, 128) results in mesh triangles with a slope of ~ 1.2 (51º). To get a slope ~1, use lighter mid Gray (153, 153, 153). Note that results are different with 16bit/32bit RGB documents; the higher range of color in 16bit/32bit can be used for finer resolution in the mesh and isn't normalized for mesh generation across the bit modes.

• Lastly, once you've created or found a cylinder model that works, delete the content on the Diffuse and Opacity maps and then export it to the collada format. Now you can open that model as a starting point, without having to build it each time.

I have a sample rgb 8bit cylinder model you can look at, but the forums has disallowed attachments. You can grab it from https://acrobat.com/#d=1lo*hYQGTj3nzaE1O-6hFA  but it might go dead at some point in the future.

2) Yes, the UV controls aren't really what you want to use for modifying the artwork to fit on the cylinder. What you want to do is go to the diffuse map and modify there. Keep the image size the same, and scale or move the layer artwork so that it lands on the cylinder where you want. The UV controls are better used on tiling texture maps.

regards,

steve

tomaugerdotcom wrote:

Right on Steve, that's exactly what I was looking for. If you could just answer two more questions I think I'll be well on my way!

1) when I've tried this technique in the past, the artwork has been scaled / stretched or otherwise distorted disproportionally as it wraps around the cylinder. What steps do you need to take to ensure that the artwork not only maps on, but scales appropriately.

2) Further to that, assuming you can get the proportions right, what do you need to do to change the scale / position of the "label" on the cylinder? The u/v controls are a little kludgy - I was hoping for a drag and drop texture repositioning tool of some sort. Also, when I reduce the scale of something (by increasing the uv scale values, the texture then tiles repeatedly, and I couldn't seem to find any options to turn repetitions off. So what would the process be if you had, say, a really large vector art file for the label and needed to scale it down so it only occupies a portion of the cylinder surface?

Again, thanks for the detailed walk-through!

Tom

[Tessa Magnuson]

This is so close to what I am trying to achieve, but I have an added complication. I am trying to wrap artwork (a diecut label) onto an irregularly shaped bottle. Is there a way to create a custom shape/mesh that I can then apply the label image to? For some reason I can't find any (helpful) info on this topic.

Any help would be greatly appreciated!

(this is the bottle, but I have my own hi-res photos I am working with: http://cache-images.pronto.com/thumb2.php?src=http%3A%2F%2Fimages.pronto.com%2Fimages%2Fproduction%2Fproducts%2F72%2F33%2Fwebsb5ea654eb973c51660282f3bcfc5-1256187707_110x300.jpg&wmax=180&hmax=180&quality=80&bgcol=FFFFFF )

[tomaugerdotcom]

Hi Tessa, the added challenge that you have is creating the actual mesh to conform to your base image. Because your bottle shape is highly rounded, you won't be able to use the cylinder technique discussed earlier in this thread. You may have more luck with a Sphere primitive shape instead of a cylinder, if your label is fairly short (in its height) and sits directly over top of the "bulge" in your bottle shape.

If you need more flexibility, you will need to look outside of Photoshop for this mesh - using 3D modeling software such as 3D Studio MAX, Maya or Blender.

The good news is that creating a bottle shape is quite straightforward in most applications. It is typically called "revolving" or "lathing". For example, http://en.wikibooks.org/wiki/Blender_3D:_Noob_to_Pro/Creating_a_Simple_Hat is a tutorial written for Blender (which is free) that walks you through the process (though they're creating a Hat instead of a Bottle).

The issue for me becomes one of - if I'm going to start modeling in another application, I might as well do the mapping, texturing and rendering there as well then, where Ill probably get more control and better quality (eg: raydiosity).

The other place you might look if you wish to stay within CS is (as has been mentioned before) Illustrator's Effect > 3D > Revolve. You won't get a photorealistic result and there's no way to export the Illustrator artwork as a valid 3D mesh in Photoshop, but this may be a good option for you if you don't want to dig into 3D.

Finally, there are lots of ways of "faking" this, as has been mentioned near the top of the post. You could use the Free Transform tool and enter Warp Distortion mode and that would let you conform your label to the shape of the bottle as well.

Best of luck!

T

[Noel Carboni]

tomaugerdotcom wrote:

Finally, there are lots of ways of "faking" this, as has been mentioned near the top of the post. You could use the Free Transform tool and enter Warp Distortion mode and that would let you conform your label to the shape of the bottle as well.

That can be surprisingly effective, especially if you add a little shading...  This isn't done very well, but you get the idea...

-Noel

[tomaugerdotcom]

Hehe. +1 for the quick mockup!

[Tessa Magnuson]

Thanks so much for the responses (and the mockup : ). I had tried using the warp transform etc. but was having trouble (i decided to make this really difficult and have really odd-angled shots)...i might end up just having to ditch those photos and stick with something more straight-forward.

Thanks again!

[SG...]

Hi Tessa,

You could also use Repoussé to generate the bottle shape, but it really depends on what you need to achieve. Repoussé wouldn't be the quickest for single still shots like this. It starts to make more sense when you need to provide multiple views, or multiple labels, or more complex photo insertion, or animation.

Here's an example, if you look closely the mesh isn't quite the right shape but with work can be better.

regards,

steve

[tomaugerdotcom]

That's pretty impressive using extrusion and bevel profiles! Seems to me that the next step for PS 13 would be to include some kind of lathe functionality using the Pen tool.

[SG...]

Hi Tom,

That is a lathe operation, not profiles. 😉

The UI just obscures this functionality, but it's there in the Repoussé dialog. Look in the Extrude section at Bend and set the X or Y angles to 360º. Also play with the 9 point reference locator and moving the Depth value to drive offset.

Enjoy!

steve

[tomaugerdotcom]

That is a lathe operation, not profiles. 😉

Wow. Next you'll be telling me to RTFM! NEVAH!.

The UI just obscures this functionality, but it's there in the Repoussé dialog. Look in the Extrude section at Bend and set the X or Y angles to 360º. Also play with the 9 point reference locator and moving the Depth value to drive offset.

Yes. Yes it does. Is there a good reason for that? It seems to me like a product manager said "we're not doing full 3D modeling in this release, so don't put an 'extrude' menu item and a 'revolve' menu item. Let's call it something sophisticated and obscure. Anyone know any French?". And then later a dev, who already had the algorithm in his back pocket turned to the UI guy and said "dude, just work in these extra parameters into the Repousse dialog box before the PM notices..."

just joshing. T

[kuzus]

Difficult

[SG...]

tomaugerdotcom wrote:

In Photoshop CS3 and earlier, taking a layer, such as a label for a wine bottle or jar, wrapping it around a 3D cylinder and compositing it to an image (of the actual jar or bottle) was almost trivially easy.

Unfortunately, with the advanced 3D tools available in CS4+, the simplicity of the Wrap > Cylinder option seems to have been lost.

Do you have pointers to this older workflow? It almost sounds like a 3rd party plug-in (ie. where is Wrap> Cylinder option?).

I'll  agree that the 3D tools have added a lot of power and complexity. I'll  see if Russell Brown has any tutorial's that speak directly to die-cut  label wrapping. Do you have experience or desire to use 3D modeling  apps, or are you hoping to just use Ps for this? I ask because the  'object' that you're wrapping labels on will often drive the ease of  success.

I can try and put together a fairly straight  forward case of using Ps 3D tools to create a 3D jar die-cut label and  post here soon.

regards,

steve

[tomaugerdotcom]

Do you have pointers to this older workflow? It almost sounds like a 3rd party plug-in (ie. where is Wrap> Cylinder option?).

Ha, yes, I'm showing my age (I've been using PS since 1990 - before "Fast Eddy" release). After a little bit of back searching, it was actually in PS9 where they had a 3D Transform Filter - I can't remember whether it was at the top of the Filters menu or inside Filters > Render. It was a very simplistic wrapping dialog box where you had onboard mapping, orbiting and camera controls, which then rendered out a static, rasterized Layer after pressing OK. Primitive but very straightforward.

[SG...]

Hi,

Found it with some searching myself. It looks like Ps stopped installing it with version 8.0/CS (came in the Goodies for a couple more versions). At least for cylinders it only wraps to the front 180º right?

I'll play with the 3D Transform plug-in some more to see what can be suggested to make the current CS5 workflow easier. If you have any thoughts feel free to write back.

Have you used any of the other 3D features yet? Have any thoughts on where it's too frustrating and you stop exploring?

Before Fast Eddy, wow! I was learning B&W Film printing back then ;-).

regards,

steve

[tomaugerdotcom]

Awesome, Steve, thanks so much for the super-detailed answer. I'll need a bit of time to digest some of the finer points relating to greyscale->mesh generation and slope.

As for how I've found the experience thus far (and this, coming from a guy who has taught SoftImage, 3DS and Maya at the college level for years), I eagerly looked into it when CS4 introduced the new tools, and quickly discarded since it seemed extremely finnicky; then more recently in CS5 I've been trying to do some experimentation and am running into roadblock after roadblock; including crashes and memory issue. I suppose 4Gb isn't sufficient these days (though my 3D apps don't seem to be complaining as long as I watch my poly counts).

My belief, being a Photoshop Old Timer, is that Photoshop has lost its way, since 5.0. I truly don't believe that 3D has a significant place in the app, considering there is so much software out there that does a better job. If rotoscoping is the use-case then I thought that simple 3D integration, such as that which Strata CX brought into Photoshop CS with their 3D plugin, is largely sufficient for most applications.

I certainly appreciate a tighter workflow integration between PS and the 3D apps, but with so much specialization and excellence out there, I think Photoshop should still strive to remain best-in-class rather than Jack-of-all-Trades. There are so many areas in which the development money could be better spent (most notably the Holy Grail, ie masking - the Refine Mask feature needs to be re-architected from the ground up) than trying to duplicate 3D functionality. I just don't think Photoshop is architected to be optimized for it.

Anyway, sorry for the rant. I'm going to keep plugging away at it because my students are clamoring for more in that area, and right now 3D is the "undiscovered country" of Photoshop, so if you're teaching that, you've got "the edge" as it were. I just wish there would be less crossover: each tool for its job.

I would be interested to learn more about the PS roadmap, to understand the use cases that are justifying this massive development investment in 3D. It may be that I'm coming at this from the wrong perspective altogether.

Thanks again Steve, for the engaging discussion, and the help thus far!

Tom

[tomaugerdotcom]

Just one more thing, now that I'm thinking of it. When we think about the evolution of Adobe's only major 3D product (Adobe Dimensions), its eventual integration into Illustrator was a natural and logical move, and its current state is pretty usable, all things told (though the mesh resolution and rendering algorithms could use with a major overhaul). The use cases (packaging and simple 3D illustration) are pretty compelling and fit well within the capabilities and mandate of the app.

Photoshop's early 3D Transform features had a lot of the same charm - simple, effective (though buggy) and a very straightforward mapping to some basic use cases (wrapping stuff on basic primitives to match objects in the scene). Although the 3D Transform was very primitive, it was very well suited to the simple uses we had for it and felt like a natural extension of the Photoshop toolkit.

T

[SG...]

Hi Tom,

Thanks for replying with your thoughts. Let me respond somewhat out of order 🙂

There are so many areas in which the development money could be better spent (most notably the Holy Grail, ie masking - the Refine Mask feature needs to be re-architected from the ground up) than trying to duplicate 3D functionality.

Are you saying the work on Refine Edge which is new in CS5 needs re-re-architecting? 😉

http://blogs.adobe.com/jnack/2010/04/video_selecting_hair_with_refine_edge_in_cs5.html

My belief, being a Photoshop Old Timer, is that Photoshop has lost its way, since 5.0. I truly don't believe that 3D has a significant place in the app, considering there is so much software out there that does a better job. If rotoscoping is the use-case then I thought that simple 3D integration, such as that which Strata CX brought into Photoshop CS with their 3D plugin, is largely sufficient for most applications...

...I'm going to keep plugging away at it because my students are clamoring for more in that area, and right now 3D is the "undiscovered country" of Photoshop, so if you're teaching that, you've got "the edge" as it were. I just wish there would be less crossover: each tool for its job.

I'm interested in this combination above. You don't feel the additional 3D features in Ps is a good fit, and yet you're being proactive to address demand from students. What are their specific use cases, if any? Do they resist your guidance to try other shoes if the ones Ps provides don't fit? What points do your students raise for why they want/need to use Ps for 3D instead of developing a tool kit that includes full 3D modeling applications?

Do you feel your advanced experience with SoftImage, 3DS, Maya is helpful to understand Ps's 3D features or does it get in the way? IOW, what benefits and hindrances do you note when trying to apply previous learning to the Ps 3D feature set? Or is it hard to answer at this point because you've been blocked due to crashing and performance issues?

I think the GPU is causing more headaches then the 4GB RAM, but that's only a quick assessment. Have you had much luck finding newer drivers?

regards,

steve

[tomaugerdotcom]

SG... wrote:

Are you saying the work on Refine Edge which is new in CS5 needs re-re-architecting? 😉

Heh, I sure am. Not that it isn't on its way. The Refine Edge in CS4 showed us where Photoshop was heading in terms of broadening its selection maniuplation toolset (ah, I remember the days of alpha channels + blur + levels + maximize / minimize). But we saw the (almost) disappearance of Extract Filter. Then its most powerful algorithm reappered in Refine Edge in CS5 and it was a Good Thing. Decontaminate Color alone is worth its weight in gold - that's a nifty little algorithm that saves us all time with the paintbrush manually un-pre-multiplying sub-pixel content.

But it still needs work - notably, it is one of the few tools yet that still feel "outside" the standard application workflow. It's its own dialog box (not a palette) and it's a one-shot, destructive effect. I appreciate that you can output to a separate layer, AND make your settings default. It still feels like it's stuck in a pre Smart Filters era.

Please note that I'm not complaining here, just pointing out that it can still get better, and mask manipulations (from my perspective, and that's only one, as the Photoshop userbase is one of the most diverse out there) are core functionality, whereas 3D is arguably not. Or let me put it this way: lots of apps do 3D, and most of them do it better than PS. But no other apps do selections better than PS and that's a fact. Play to your strengths is all I'm saying.

I'm interested in this combination above. You don't feel the additional 3D features in Ps is a good fit, and yet you're being proactive to address demand from students. What are their specific use cases, if any? Do they resist your guidance to try other shoes if the ones Ps provides don't fit? What points do your students raise for why they want/need to use Ps for 3D instead of developing a tool kit that includes full 3D modeling applications?

You're right, my comment warrants clarification. The demand from students is not for 3D functionality per se. The demand for students is that you teach them the latest and the greatest (regardless of what industry actually expects out of a graduate's skillset), because they saw it on Adobe TV, or they read about it in some product review. I spend 24 hours of class time on colour correction (and that's scratching the surface), but about 5 minutes on Content Aware Fill. And I wouldn't even touch it if it weren't for the inevitable savvy student that pipes up in class when I talk about the Clone Tool and says "but isn't that replaced by Content Aware Fill?". So you have to touch on it.

One of the strange new realities of teaching Photoshop post PS5.5 is telling students "just because it's in the menu doesn't mean you should use it." This was first true with things like magnetic lasso, or art history brush, or the spot healing brush tool (though very specific cases can be made for each of them - I'm aware of that) and the tradition now continues on with Content Aware Fill / Scale, Mini-Bridge (though I'm sure this is just a personal opinion, and a good case can be made for it in your workflow on a very fast computer), and IMO, 3D.

So I don't necessarily see a contradiction. Students who are fairly new to the industry are hungry for the "cool factor" (not to mention the "easy way out"), and it's a constant balancing act designing curriculum that emphasizes best practices, while still covering all the bells and whistles (with the appropriate cautionary tales). If I were still teaching Advanced Photoshop to PS pros and they were clamoring for more 3D, that would be a whole other story. I'll admit that I may be out of the loop there, though, which is why I am still eager to hear from you who you think the target demographic that wants true 3D in Photoshop is.

When I was teaching to pros, they were mostly pro Photographers (think Yuri Dojc) or Architects. I can certainly see the latter being curious about 3D in Photoshop, but those that are technical enough are now all versed in the Autodesk suite, so they would be doing texture generation and compositing in PS at most. Photographers are much less likely to be even remotely interested in 3D within Photoshop, but that's probably not the audience you were building it in for anyway.

Do you feel your advanced experience with SoftImage, 3DS, Maya is helpful to understand Ps's 3D features or does it get in the way? IOW, what benefits and hindrances do you note when trying to apply previous learning to the Ps 3D feature set? Or is it hard to answer at this point because you've been blocked due to crashing and performance issues?

Without a shadow of a doubt, my background was essential to being able to pick this up in PS without the manual. I think I would feel completely lost if I didn't have 10+ years of 3D modeling, texturing and animation behind me (note that I don't consider myself a pro in 3D - just an experienced amateur). Possibly even WITH the manual.

3D is a wonderful world full of really cool technical details that merge the artistic with physics. Photoshop has bitten off only one branch of that, namely shaders, but that is in itself a vast subject. The amount of background knowledge you need to have under your belt is vast - what's the difference between diffuse and ambient colour? what's a specular highlight, and how do you control glossiness? what's a "normal"? hell - what's a "lightsource", and what are their different modes?

I just covered the basics of 3D using repousse and height mapping the other day in a class of twenty-odd 20-30 year olds: I spent 2.5 hours with the students (we had a lot of fun mind you) and didn't even get to texturing at all. Not a single mention of maps, colour or anything. There's too much other material to cover - navigation, object-level vs mesh-level manipulations (ask me my opinion on the usability of your implementation of that feature some time, you'll have another page to read!), then TSR, then lightsources, camera views, ground planes, raytracing, merge 3D layers (you'll have to explain the logic behind having separate 3D "scenes" co-existing within the same document some time).

Bottom line: adding 3D to Photoshop seems like you've put a tiny little door in the bottom of the vast, vast world of Photoimaging. When you open that door and look through you discover that there's this other huge huge world of 3D modeling just on the other side. In order to make ANY use of the 3D features, such as they are now (as opposed to the previous incarnations - including Vanishing Point, which had some very interesting potential) you now have to jump into that world and start to get your bearings.

I think the GPU is causing more headaches then the 4GB RAM, but that's only a quick assessment. Have you had much luck finding newer drivers?

You are almost certainly right. I need to get back to Chris on another thread and let him know that my drivers WERE, in fact, embarassingly out of date. Updating them didn't solve all my problems, but a few have gone away thankfully. I have yet to push the limits of 3D on my personal box again, but will certainly circle back with you if I notice some improvement.

Thanks Steve, always a pleasure talking about this stuff. I am about as passionnate a guy about Photoshop as you'll meet, but because I spend 90% of my time in industry using Photoshop as a production and creative tool (I only teach 1 day a week these days), you'll probably find I'm quite conservative. Heck, much of our agency is still running the full CS3 - there just hasn't been enough compelling reasons to upgrade across the entire organization yet.