Elements of Technical Illustration.

[Synthetik]

I'm starting this forum post as a place to reply to craig discussion and my response to his reply in the 'On Artistic Series 2' forum post. I wanted to start a separate discussion because as i said as the developer of Studio Artist i really want to get inside the head of what is going through a commercial illustrator's head when they create what i would consider a brilliant commercial illustration sketch. I was using Joe Ciardiello's work as an example of the kind of stuff i'm talking about. So what interests me about this style of illustration? The artist is able to convey the essence of the original source photo with a relatively small number of ink strokes. In Joe's illustrations he's making use of characature so whatever that essence he sees in the original image is super conveyed, presented larger than life in the actual sketch. So what i call 'compression of visual information' is actually being used as a mental sharpening device in that it forces the viewer to perceive certain things and ignore all the other stuff going on in the original photo. I really want this discussion to focus on technical elements of drawing. I fully believe that someone like Joe get 'into the zone' when knocking off one of these illustrations, but that does absolutely nothing to help me understand what is going on technically in a sketch like this to try and add tools that recreate those technical decisions in software. We have a number of other forum topics devoted to losing the fixation on technique, and those are very valuable and relavent discussions. But i would really like to not get into that side of the whole thing in this discussion if that's ok with everyone and really focus it on technical aspects of creating an illustration like the example i posted. I do think that some of the things going on in the new Sketch features in v4 are a first step towards moving in this direction. There's a whole new level of visual analysis technology i've been working on that could and will be added to Studio Artist in the future. There are visual perception modules underlying the Studio Artist auto drawing engine now but i really want to kick them up a notch I really see the kind of stuff Craig does for a living as a commercial illustrator as falling into this kind of thing. Not being a technical illustrator myself, i want to understand the mental and physical process that goes on in creating illustrations like this in excruciating detail. SO i hope that some of you who are into this kind of thing can help give me some insight. I see that insight as getting a better understanding of the kind of visual analysis going through your head, better understanding of technical aspects of drawing you are using to convey that visual analysis, tricks of the trade, etc.

[craig deeley]

Wow!
What a rich topic.
A lot has been written on the subject.
Boiling it down is a tough task. Making it work as a process in SA sounds even tougher!
But its also one of the most interesting subjects an artist can get on to.

I follow that well worn path of simplification and exaggeration.
You will see caricature artists at large theme parks using this - they are given basic training in the tricks. Basic cartooning tricks with the twist of drawing from a model rather than imagination or accumulated knowledge.
The same tricks are what animation artists (2D) are trained in.

Schools often use the same basic tools - that is techniques and procedures for training the eye. Particularly those that follow the kind of basics that Kimon Nicolaides is well known for writing up:
From Wikipedia:
He advocated a three-pronged approach to learning to draw:
1. slow and meticulous contour drawing,
2. free and rapid gesture drawing, and
3. vigorous tonal drawings of weight or mass.


Techniques are a little easier to discuss than intentions.
They are really just standard tools for depicting something.
Because I use SA to generate material that has to make its way into the commercial markets - the standards are something I have to stick close to - if not totally imitate.

I will try to talk thru a little of the mental activity that goes into a caricature following this...

----
----

In an aside:
My suspicion is that some technical aspects of drawing can be restated as recapitulations of sensory information being received/interpreted.
The eye has receptors for different input:
Rods and cones
Some see specific colors, some see black only, some see clumps, some see details...
We interpret whats going on outside by what it stimulates in the eye.
A lot of the techniques of an artist seem to recreate the kinds of input the eye receives/interprets.

That would be another subject on the mechanisms of perception...


Craig

[Synthetik]

The nerdy scientific viewpoint of caricature is that it is an exaggeration of specific features that deviate from a standard norm. So for facial caricature you could start by constructing a standard face that you could think of as the average of all existing faces mixed together. Then you measure the variation of a real face from that standard face. Then you extend those measured variations to construct the caricature.

So, using the Studio Artist tools you would construct a bezier warp from the standard face to your specific face features by drawing bezier curves on the appropriate features in the 2 images, record that as a warp context paseq action step, and then extend the specified transformation to your real face image to generate a extended caricature image you would then derive the automatic sketch off of.

You could also think of it in terms of morphing if that makes more sense than my garbled explanation above. You morph the image to extend deviations from a standard normal image.

I'm real curious to hear your upcoming explanation of the mental activity that goes into a caricature from the standpoint of someone who actually draws it.

[Jean Detheux]

Before the eye sees, the mind intends, and it is that intentionality that guides what is being looked for and how (and thus seen).

Again, "The Hidden Dimension" by Edward T. Hall shows magnificent examples of that, as for example, of how the eye scans an image based on different "instructions" given to the viewer (how old, rich, sexy and all is the woman in the picture, each one of those giving birth to a totally different pattern of scanning, building a totally different "representation").

"Perception is constitutive," we can't see something (or rather, "some thing") if we do not intend to see it, if we do not already know what to look for (just think of your being in front of a still-life or landscape or whatever, and loading red paint onto your brush; what do you see when you take your eyes from the palette and set them on your motif?).

"What do I see before knowing what it is I am looking at?" was the topic of one of my many lectures, it's even the subtitle of one of my web sites.

Art is also, if not primarily, the investigation of our perception of the world before we make sense of it, or, as been said by Ronald Hayman in the Yale Review (Spring 1980): "The mature Cézanne had no designs on the field of vision except to uncover the designs he saw in it. It is this suspension of will power that gives him admission to the undifferentiated world which precedes knowledge, to Eden as it was before Adam conferred separating names on each form of vegetal and mineral growth."

Cézanne was so in tune with his "here and now," he had been known to quit a painting because his head had moved a bit and he had "lost his motif."
Had he worked on the societal model, it would not have mattered, the "out there" woudl have remained constant, regardless of where/how his head would have been.

I can see the worth of some kind of simplification of images as called for by the commercial world, but that does not make the results of that simplification a reality connected to the world of genuine experience.

Genuine experience lives in the world of the particular, the "how do I see?", while commercial art serves the world of the average, driven by societally acceptable model(s), the "how we (should) see."

[liveart]

The eye is essentially a sensory input device for the neural net that is our brain. Despite some minimal signal processing in the aforementioned rod and cones, the eye does not "see" at all. So it might be possible for an artist to endeavor to show what the eye sees without thought but the only accurate way to represent that would be something like an blip on an oscilliscope. Everything else is a representation of our individual intrepretations of this input.

But that does not help much with how a visual perception module underlying Studio Artist might be constructed. I do believe that one issue that would need to be examined is the fact that we are Binocular.

The brain constructs and anticipates, "intends" in Jean's termonology, based upon the fact that we use stereo input to understand an object's 3d features.

This is relevant even to the 2d representations in caricature. If you look at the simplification used by the caricature artist, it invariably uses the topological features as a starting point. (not the same as edge detecting). A bezier curve that could follow topology would be wonderful, then add the ability to smooth or warp those curves and you'd be well underway.

Of course this leaves the problem of intrepreting 3d features from a 2d image. There is some considerable research on this and a few dedicated packages.

This does not mean a statistical analysis of the image with the overlay of topological bezier curves will provide an automated way to do an effective caricature. The human element of the caricture is to know which features to exaggerate. Not just which features but which type of "warp". A nose is easy to pick on (pun ignored). But do you make it bigger or smaller? This might depend on the purpose of the drawing.. a scathing political cartoonist might make a different choice than a caricaturist doing a party favor. How about personal blemishes? a wart, enlarge it and add detail to this area to make a point.. or a birthmark.. Gorachev's for example.. These are the elements that are probably beyond a perception module.

Beyond the caricature, Illustration is generally a process of simplification and detail recovery. The choices of where to add back the detail vary again with the "intention" of the artist. BuZZ is a Photoshop plug in that attempts to do do "intelligent" simplificaton. The core of this is a filter that removes "unwanted" detail "without blurring, distortion, loss of focus, colour, and edges." Defining unwanted detail is probably the more significant issue.

I posted somework based on a study of perception which reported the attention to detail in an image is culturally biased. Specifically "western culture" subjects in this study looked for and recalled greater detail in the center of an image, whereas "eastern culture" (primarily chinese if I remember the study) look for and attend to detail at the periphery of an image.

So I can see where the process can be described as simplication and detail recovery, without that meaning one can implement that process into a visual module..

[Synthetik]

The existing visual modeling buried deep within Studio Artist involves modeling the visual cortex, not the rods and cones of the eye. So for the technically inclined we try to model visual areas V1 on through to the IT cortex. There are many computational trick simplifications to pull this off, and our approach to things like V1 or V4 are gross simplifications of everything going on in there. It's mostly pre-attentive vision, although you can configure the paint synthesizer to build presets to emulate attentive saccadic tracking.

Your comments on 3D perception in 2D line drawing are right on the mark in some respects. The new v4 Sketch features do not take advantage of that kind of visual modeling, so it's an obvious area to look at to improve this kind of auto drawing in the future. I actually do know quite a bit about how to model this and hope to add this kind of thing into Studio Artist after v4 is released. Knowledge of what is going on at the level of object recognition in the IT cortex has also increased quite a bit since Studio Artist was first released, so again i'd love to beef up that whole part of the underlying structure of the program after v4 is released.

i realize the neuroscience talk blows by most people on this forum, so moving on, you talked about using intelligent simplification via image processing. There's a lot you can do using the Studio Artist ip ops to manipulate a source image prior to generating an automatic sketch effect to facilitate this. The Smooth or Simplify ip ops are obvious places to start. again, the approach one would have to take with the existing program is to build PASeq's that incorporate this kind of thinking into the overall approach to building a sketch.

There's also a lot you could do with hand selection of background areas followed by blur operations on the masked background areas to simplify an image. It would be nice to see that whole process become more automatic, again features i would like to add in the future. In some sense, you can use ip ops like Smart Contrast routed to the current selection region to build masks that emulate that kind of thing as opposed to doing it by hand. There are a number of PASeqs that are provided in the 3.5 factory set that work that way. A selection mask is automatically constructed with some ip op, and then used to auto mask painting. And you can always invert the mask and then do something different with the inverse selection.

Your comment about cultural bias in attention to detail is fascinating. Do you have a link to that?

[Jean Detheux]

John,

It's not "just" attention to details that is conditioned culturally, it is above all the perception/constitution of the whole, that which we "draw" details from.

If interested in how different cultures constitute and use social space differently (and this links to "attention to detail"), that book I keep mentioning here is pure gold: "The Hidden Dimension" by Edward T. Hall.

[liveart]

I do not have the link to the original article. The work was by Nisbett. I did a quick search and found this: http://www.ijdesign.org/ojs/index.php/IJDesign/article/view/267/163 which while slightly sideways to the original contains enough info to send you on a discovery path.

[Jean Detheux]

Yup.
And Edward T. Hall has two books listed in that paper's references.

[craig deeley]

I will walk thru some of what goes on for me when creating a character in a very "cartoon" style:
http://en.wikipedia.org/wiki/Cartoon

Starting from some form of established figure (personage, thing) I do several things in the process of creating a "cartoon" version.

I have a good look at that figure. But not a long, long look. I look straight and look sideways - that is I let it drift out of direct view as well - I note details and get its impression... (I will recheck details and impressions if I continue to look for information about the figure - but first glance is like the above)

I note and then re-categorize - mostly unconsciously and VERY quickly - what "vibe" is coming from the figure.

Typical things that are clear in a seconds glance:

Likable, approachable
Open/soft/light(bright) = round lines minimally shadowed - smiles in eyes

Standoffish, dangerous
Closed/angular/dark = straight lines and shadows = 45 degree facial lines
(These are also the cliched attributes of "Female" and "Male")

These same aspects can be attributes of objects... everyone and everything is made of mixes of these "lines"
Character designers and animators are taught to mix a little straight with a little round to create a more interesting character/personality!

----

I take the gut feeling and associated lines and think about the space the thing occupies and its "action" lines... Basically I start to recast the thing as volumes and action vectors(where it is going)

When learning to draw - for instance using basics for translating what is seen to a surface - like Kimon Nicolaides steps posted above... You might pick and choose an approach for constructing a figure. I have long ago mixed the steps... touching contour or volumes or action lines and mixing them - looking to define the form I am looking for.

Clearly I want to remove the figure from its ground while developing the figure...
Basically just ignoring the ground for the moment. Tho the ground is often implied by how the figure rests on it... The figures arrangement of parts.

I have nothing but a blank paper (or application "canvas", stage etc...)

Now I am going to start "building up" from "nothing"...

When starting to draw - I might make a fast action stroke to define the vertical "spine" of a figure... And immediately ground it with a horizontal "foot" or expand it with a "chest" (center) or a "head" (top)
And then fray my lines out from foot or chest along lines of action to extremities.
The spine and chest basically become the hub... The "spine" typically carries the line type and personality of the figure (straight and or curved)

I would typically start bulking out the shape with masses at this point - using a contour mass rather than a filled area (but occasionally the reverse - filling only - like drawing a shadow) - using standard shapes - round tubelike shapes or angular shapes (balls and boxes)

Because I practice this every time I draw... I may only mentally perform any number of the preliminary steps to building up - and jump directly to a stage anywhere along in the process. If I skip before I know where I am going I might end up with something that totally stalls the process - for instance skipping to some subtle detail before defining action lines that the detail really aught to fall into place along.

Honestly - In actual practice - I try not to follow a defined step by step process... Like counting the steps in a waltz instead of dancing... I skip around looking for serendipity... And occasionally totally miss the mark - trying not to stifle the "creative energy"... I usually go thru several false starts and several attempts at the same subject... My early takes are often shape or action tests... Looking for what pleases... I know I will go thru several attempts before hitting on the shapes and details I like...
Anticipating this is another matter.

---

Here there's room to think about:

- starting from nothing (blank canvas)

- Kinds of lines (round or straight) having recognizable associations - that most people share.

- Examining mass and implications of "motion"

- Indicating mass and motion with a "spine"...

- putting "flesh" on the "spine"...

I am going to stop here and follow in another post - exaggerating and eliminating things.

C-

[liveart]

So I think that a very good discussion has been posted on the how the mind directs visual perception.

One set of examples of this can be found in the realm of optical illusion. More or less my definition the mind greating misdirection. Here is a link to a famous one that still amazes me. (I have even used this premise in both sculpture and jewelry designs).

https://www.youtube.com/watch?v=QbKw0_v2clo&eurl=http%3A%2F%2Fww...