Tracing Chair.jpg (Chapter 12)

[440bx]

Hi Gary,

Me again! This time picking on the lawn chair.

I used the settings indicated in the book but, the result I get isn't quite close enough to what you got. I should mention that the differences don't really matter much (artistically that is) but, I'd still like to know how you got your chair to look the way it does.

Specifically, my chair ends up with lots of little transparent "islands" inside it. When I zoom in my chair, the islands become larger thus more visible. Your chair doesn't have any islands even though it is possible to tell that there are very tiny transparent areas between the outlines of its component shapes. The amazing thing in your chair is that when I zoom in, those tiny islands disappear (which is great) instead of getting larger and more visible as in mine (which is not great). This behavior alone got me into the mode of "how did he do it ?"

When I zoom into your chair, I get the impression that the smoothing was not set to zero as indicated in the book. I get that impression because the shapes that make up my chair have sharper edges whereas in yours the shapes are clearly more rounded.

I have enough chairs to build a movie theater in N.Y but none are close enough to yours for me to be satisfied.

I found that by setting the smoothing to somewhere around 20 and increasing the number of passes, I get closer to your chair but not close enough.

The attached document shows the bitmap I started with, the result of my trace (one of them ) and a chair I stole from your lawn

The differences I mentioned are particularly apparent in the armrest and its shadow.

Thanks for looking at the stuff.

John.

[Gare]

Well, I confess that the results I achieved were partially Dumb Luck, John.

The point of the example is that a photograph that looks highly posterized is the best candidate for auto-tracing and achieving a result that looks photographic. Because a softly shaded photo is going to wind up looking posterized no matter which way you slice it—so the process begins by picking out a very contrasty subject, or forcing it that way by posterizing the photo before auto-tracing it.

By "islands", you mean gaps between shapes, right? Trial and error is the only remedy, and then after that, some manual editing. You can also put a shape behind adjacent shapes of a similar color to quickly close any tell-tale gaps.

Here's the interesting thing: zoom in and out of a trace. Sometimes you'll see gaps and sometimes you won't. Why? Xara Xtreme was designed to anti-alias onscreen: when your viewing resolution hits a fractional integer, you'll see edges, and then at whole number viewing resolutions, you won't.

Aesthetically? Your auto-trace looks as good if not better than mine, John. It's what the product looks like, and not the settings! All you need to do is delete a lot of the trash around the edges and you're all set to group then distort the chair. If you traced a little closer to the edges before making a bitmap copy, all the green edges (or most of them) would go away in the trace.

But it's the creative process, and how one does it, not the chair. Try auto-tracing a photo of a statue, or a tree on a sunny day, something that shows a lot of contrast. But before you do that, think of a concept: what will you ultimately do with the auto-trace?

Auto-traces can't live in an inspirational vacuum .

My Best,

Gary

[440bx]

Well, I confess that the results I achieved were partially Dumb Luck, John.

I'm sure I can repeat the Dumb part, the Luck part has proven more difficult (as it usually is)

You can also put a shape behind adjacent shapes of a similar color to quickly close any tell-tale gaps.

I thought about that in one of my many tries. I placed an untraced copy of the chair behind the traced copy, that took care of the islands/gaps. This worked very well as far as the colors that were used to fill the gaps but, pixelation was noticeable when zooming in those gaps. In this case the pixelation doesn't matter because the chair does not need to ever be very large.

I was going to attach a file to show an effect I found unusual at first, in the process of doing that, I learned how important the "never smooth" setting has. Unchecking it got rid of the pixelation (it activated the Xtreme antialiasing engine duh! - I knew I'd be able to replicate the Dumb part ) I think that setting may influence the result of the trace as well. I'm going to play some more with it to get a better feel of how it all works together.

when your viewing resolution hits a fractional integer, you'll see edges, and then at whole number viewing resolutions, you won't.

That explains both the nature of the gaps in your trace and why they disappear. Perfect, thank you. It also explains why the gaps don't disappear in my trace.

... what will you ultimately do with the auto-trace?

I think that, generally speaking, I would use it as a starting point. Enhance it from there. The Globe logo is a good example.

Thanks Gary, this helps put the pieces together.

John.

[Gare]

Here is a work exercise above and beyond the Guide, and a few tips for making an honest-to-gosh illustration out of an auto-trace. Tracing alone don't make it "Art".

However, tracing is not "cheating". In fact, it's a good way to develop eye-hand skills and better learn the relationship between colors in adjoining image areas. Photorealism? The colors are duller than your eye alone tells you, and areas of contrast aren't really where you think you see them. I trace all the time, sometimes auto-tracing, other times manually tracing. It's for the sake of accuracy, and for saving time...I heard somewhere once that a computer is a time-saving device. Think about it: would you really want to drive a car whose engine parts were designed by eye alone? Except for Toyotas, most auto parts are traced using one means or another, as is medical equipment and in any field when you need fidelity. And that was only a semi-deserved snark about Toyotas.

1. The pawn and rook scene in the image below was first created by auto-tracing (at left). At right is a hand-edited version. What I did was set up a nudge distance far enough to move objects into the clear to the right, and choose "good" objects, duplicated them (Ctrl+K), and nudged them for my final drawing. They aren't going to perfectly align if you open the Xara file and try this because I did a significant amount of hand-editing and scaling.

2. The image (attached) is a good, but not terrific candidate for auto-tracing. Although there are some nice sharp highlights, the chess pieces are smooth and the soft tent lighting causes some gradual light fall-off. Because auto-tracing segments graduated areas, you wind up with some unwanted posterization. The best way to correct this is to import the image, examine it and see how a shape can replace several shape segments through the use of transparency, multi-stage gradients, and feathering.

3. Most importantly, you don't try to duplicate an image. What does that say of your art? A: nothing. It tells the viewer you have great technical skills but not the vision it takes to re-interpret what you see in a photo, or in life.

Do you copy?

My Best,

Gary

[440bx]

Wow! That's beautiful.

Thanks, I will work with that image and attempt to recreate the final result (for the sake of developing the skills)

Back to reading and learning,

John.

[Gare]

Do I know how to pick an image to cater to a specific audience or what?



-g-

[440bx]

Do I know how to pick an image to cater to a specific audience or what?

To say the least great choice.