Self Critique Techniques

[fiona]

I am not very good at looking at my own artwork and deciding if it is working, or not. Thus I have learned some tools along the way that I apply to artwork to test to see if my "gut instincts" are matching the rules of design.  These are some techniques I have used for "testing" artwork...both my own artwork, and others when they have asked for feedback.  I hope these techniques are helpful to others to analyze their work!

Test 1.
a.Technique:"The multi-resource color comparison" 
b.  Question: Are the colors of the artwork matching my vision?  Is light of different colors interacting the way I want it to? 
c.   How to: I find multiple resources that have some one or more elements of the color composition I am trying to create with the mood, lighting, and color in the scene.  As I find them, I place them around the borders of a large canvas on a 50% grey background.   Sometimes I number them, and write a note to myself about what aspect of the composition I want to emulate.  I try to arrange them into groupings that have similar aspects that I like.  In the middle of this large canvas, I leave room for before, and current images, with a third empty spot for the "next render".  As I make new renders, I add them to the empty slot at the right of the current render, and ask myself.  "Is this an improvement, is it coming closer to my goal?  I look at the whole picture, and either keep the new render or discard it.  IF I keep it, I discard the image to the left, and slide the two remaining images over a slot, making a new empty slot to try again... Another version of this technique is have my first image be "where I began" so I can see how far I have come and if it is overall getting better, as a process.  This second method sometimes keeps me from heading down a wrong path.
d.  Example1 :


e. Example 2 This is another example, with examples of the types of notes that would be written.  This time the notes are on the image, since the critique was done for a friend.


Test 2.
a. Technique "The 50% grey convert to B+W challenge"
b. Questions. " Does my image have enough contrast to be readable?  Are details of the textures overwhelming the work, or are there too few details?  Is there variety? Are the lights and darks accentuating the correct parts of the image?
c. How to:  I create an image background that is 50% grey, at three times my image width +200 pixels, and the height + 100 pixels.   I then go into photoshop and choose the "mode" to be "greyscale".  Note.. this is different than convert to black and white..  I want to remove all color information, not allow myself to "cheat" and increase the contrast, as the options in "convert to black and white" would allow me to do!  I convert my image to greyscale mode, and copy and paste it onto the 50% grey canvas.  I look at it critically.  Does it have enough contrast?  I look at the lights and darks.  There should be a very, very few little spots of nearly 100% white, the brights, and very few spots of nearly 100% black.  Depending on the mood, the middle tones of the image should be at 50% grey.  The details of the image should be at 25% grey for the lights and 75% grey for the darks.  The image should have a balance of both light and dark details. 

I then work on how the light hits various parts of the image.  In the attic girl image, I had a very bright moonlit window, and a very bright lantern.  Between the two was a little girl who was the focus of the image.  I was working from two different renders.... one with the chin tilted further back and more "haze" and one with the chin more towards the light and less haze.  Undecided which was the best route to take, I was trying to get the best of both images. Doing the black and white comparison gave me a very important piece of information that I would have missed otherwise, and gave me the impetus that the left image was the one to move forward:  The light caught the eyes.  I liked certain aspects of the image on the right, and so I kept it as a reference as I moved the left image forward, trying to capture more of the light from the moon on the pillow, for example.  Continuing to check how the image read in black and white let me keep control of keeping the focus on the eyes.

d. Example:



Test 3:  Posing Accuracy
a.1 Technique 1:  Use a photo reference to copy a pose exactly!
b.1 Question: Are my poses lifelike and natural, or stiff?   Do they match references?
c.1  How to:  There are many ways to use photographic references for posing tools in creating computer generated art. One method, when it is important to create a pose exactly, such as when recreating a dance, is to take an image of the pose to be achieved directly into the work plane.  This is especially effective if there are both frontal and profile views of the pose.  (This can also be used to recreate facial features!).  To do so, I create a simple plane primitive, with the photo to be used in the diffuse of the surfaces tab, and scale my rectangle to the same proportions as the image so as to minimize distortion.  Then, I place the primitive at the zero point, angled to face front.  I create a camera that is also the same dimensions so as to capture the primitive on the full screen. I add a character, and move the primitive up the y axis til it bisects the hip bone.  I scale the figure to match the photo, and go to work, recreating the angles of the limbs from the center, or hip bone, out.  If a profile view of the image is available, I place this also at the zero point, and depending on its orientation, rotate it on the y, or vertical axis, either 90 degrees positive or negative.  I create a camera for this angle.  I work alternatively from each photo angle.  It may be necessary to move plane forward or backward from the zero point... it should always bisect the area being worked on,  the distance from the camera should take care of scale issues.  Then, when I think I have come close, I take a render.  I overlay reference over the render, and check my progress.  I continue until I have come as close as the figure will allow me to. 
d.1: Example: top image, reference; middle, render; bottom, overlay


a.2 Technique 2: Use a photo reference to judge the accuracy of a rendered pose- centerline, limb angles, negative space
b.2 Question:  How can I tell if a pose is accurate if my figures have different shapes?
c.2 How to:  When the characters are not identical in a pose, overlaying them will not work.  Another method to discern if the pose is accurate to the photograph is to check limb angles, centerlines, and negative space.  In the first image in the example, when drawing a line from the endpoints of each color-coded bone, the lines of each  bone in the render on the left, should match the angle of the bone on the reference, to the right.  The render and the reference look very similar until the lines of the bones are drawn.  Then it becomes apparent that the render is not at the same angle on the hill, it is further over the hill.  This is forcing the kneeling figure into a more upright position.  Secondly, looking at image 2, if the pose is accurate, the shape created by the limbs in relation to the body, called the "negative space" should create a shape that is similar to the negative space in the reference. Checking the green under the arms and fuschia under the legs... are the shapes the same between the render on the left and the reference on the right?    Lastly, the figures should be the same distance apart, but drawing the blue center-lines helps the eye to see that they are further apart in the render than in the reference.
d.2 Example:


Test 4:  Composition
a. Technique:  Using the "Rule of Thirds" and the "Golden Rule"
b.  Question:  Does my composition make use of the Rule of Thirds "Hot Spots" or the Golden Rule's 3:2 spiral to create a composition that creates focal points while allowing the eye to move around the page?  Or is my image overly centered and static? Or are the figures too distant, and lack a sense of urgency in their interaction?
c.  How to: When setting up a scene, visualize the camera plane divided into a grid with nine equal squares, or divided into 3x2 rectangles that get progressively smaller to form a fibonacci spiral.  Try to "cross the eyes" either with the crosshairs of the grid, or the spiral.  The example given below was a critique given to a friend.  He incorrectly believed he was using the fibonacci spiral to create his composition.  However, although he did use a spiral, because it was distorted from the 3:2 ratio of the fibonacci spiral, the two figures were too far apart to be seen as interacting in a dynamic manner.  In the set of images to the left, as much as possible, I used the recomposition tool to move the figures into a "Rule of Thirds" scenario, trying to "cross their eyes" with the "hot spot" dots of a grid overlay.  The set of images on the right are the same idea, trying to cross their eyes with the corrected fibonacci spiral overlay. 
d: Example.

[sanbie]

Wow...very, very informative    

[fiona]

I am happy to find out you can texture paint in blender...but can you copy onto the brush from a photo and reproduce the texture on the model... like painting on skin? I know there were old versions of Blender that did this, but they took it out.  Check your new version to see if it does it.  If so I will come kiss you,  and hope you don't remember   I asked Chris at Hivewire the same question two days ago and he said he didnt know a program that was free but hes not a texture guy. He told me to partner with their skin guy if I couldnt figure out skins for Dusky.  I am not allowed to call him dusk ____.  He has now been renamed Manny for Dusk, short for Almanzo.  I would have used "Manly" for Almonzo but that is likely copywrited to the Little House folks. Maybe you should set up classes, Professor Hedd!  You know a lot more than many many people!

I am glad to learn that Blender has these options on the camera.  It probably has contrast adjustment on the camera too.  *hugs*  Daz, pretty simple cameras.  Daz... designed for newbies to 3d with a fairly low ceiling of capabilites.  Blender... designed by rogue freeware affectionados who want to balk the money grabbing big guy programs and be able to do it all.  Widely different user market... widely different capabilities and ease of use. 

I know most vendors dont bake maps.... its my rant that baking em is the better way... and a girl can wish and hope.  Just cuz a lotta things suck monkey balls dont mean I gotta like it.  "Manny" is gonna have baked maps.  And some people don't even know there's a different, other way!  That texture can be a reflection of something true to life, and true to a very detailed model that someone invested a good deal of time making.  And I didn't even talk about dirt maps in my map tut.  Or the ways that blender affectionados tell you to bake 3  different layers of "dermis" for lifelike skin..the thin outside layer, the ugly yellow fatty layer, and the raw meat colored flesh layer.  I was holding back! 

I agree 1000% that specular maps are sucky and a waste of time... but most of the time in Daz, if someone is lucky enough to get mats info besides just a color and a bump map, they are going to get a specular, because its so easy to rip them out in 2d by reversing your bump.  and well- I would rather use reflection than specular at all.  But if you are going to use it.. at least stick something in the slot relevant to the mats you are working on.  Most times it is used as a proceedural to save render time because good mats  take longer to render.  Something about the program having to refer to the mats each time it wants to make a pixel instead of a formula...

Lots of vendors painting their maps are not doing so in 3d but in 2d.  Particularly newer folks who cant afford to lay out a few hundred bucks or more for a program.. or older folks who are from the days before these things were possible and haven't converted.  Painting mats in 2d for use in 3d is wicked hard and I give those who can do it serious props. 

Lots of people, I am told, skip putting clothes and hair on their characters in 3d at all, and just paint them in in postwork.  That sounds like fun, but for me, I am a builder, I don't really like or want to hand paint in 2d... not until I can paint directly on my touch screen, and maybe not even then.  I have a pen and tablet, and learning to control it is seriously frustrating to me... so props to whose who can and do. I suppose down the road a ways here, practicing is going to become necessary if I am going to have control when transfering skins from pictures to a model, but thats still way easier than drawing free form with a pen and tablet.

Blender and from what little I have seen of poser, for materials you get nodes and plug channel "a" into channel "b" and kind of play connect the dots and get to tell what percent of each to use.  In daz, you get a surfaces tab, which houses a "shader" with map slots, color sliders, and strength dials.  Yes, you can "build your own shader" which seems to be exactly the same as what other programs do with nodes, except they call them bricks.  You choose which bricks you want and connect them all up.... sound familiar?  But the bricks all have weird names and it is going to take sitting down with their dictionary to figure out what the frck they do.  I will write a translation tut when I know 

In Daz, anything beyond the most basic "pick character, pose it, light it, and take a picture" is not a system that is created to be easy to use.  Remember, Daz has a different marketing scenario than Poser, and Blender (which has no marketing scenario other than "give us money we are cool and free").  Poser you pay for and get 7mg of content... big initial layout, but you get support, you get how tos. you get content, you get a program that is does many things with a little learning investment.  Daz is free and intended for completely new 3-d users, many of whom are content to stay clueless... and Daz is happy to cater to that market.

But then when a Daz user wants more... what do they do?  DAZ HAS A SOLUTION! Buy an addon from Daz!   Anything beyond the most basic, I personally believe Daz deliberately makes hard to do on our own without reverting to someone else's presets.  This way they can market user guides, market shortcut addons, market "different" characters which a person could make on their own if they wanted to invest the time.  Daz is intended to be very simple to learn the basics, and hard to go beyond that, so they can nickle and dime money out of those not wily enough to know better, or too uninvested to care, or too time lacking to be able to go deeper than making basic pretty pictures.

There is another that Daz markets to...those who have money to spend and enjoy spending it on pixels.  There is absolutely nothing wrong with that.  I would spend money on pixels, moreso than I do,  but I lived for a very long time with absolutely no discretionary income.  If I have extra money now, I feel guilty... like I should either give it to someone who needs it, or invest it in learning, so that other people can learn from my learning, and the world become smarter from the money. I have this relationship with the concept of money that is complicated.  I am not advocating it for anyone! It stresses me out. 

So why am I still using Daz?  Because in the beginning, like Blender, it was free, and I didn't realize the nickle and dime money bleeding scenario that lets them stay afloat.  Then, having fallen for the nickle and dime, I had spent more on Daz than I would have on a  poser layout for a better program, or better yet continue to save for things like Blacksmith, etc.  I was not in a position to really pay again for 3-d.  Now, I have $$ invested in Daz and Poser-ready characters, which I would need to re-rig to make useable in Blender, etc. 

I am doing more and more in Blender but still using the Daz software because after 5 months its what I am beginning to know and I am committed to unlocking its doors so I can avoid giving them more dough, and soon will add Poser to my repertoire.  I am considering vendoring some of what I make in addition to freebies, and to vendor I need to know Poser as well as Daz. I dont want to vendor to make money, per se.  But because at some point to keep learning, which is the most fun part of anything for me, you become a creator of content, in some way.  I will give some of it back to the community that taught me.  But I would love to get back the investment, on some level, as well.  (The things I spend months developing and perfecting will go to market, the testers, to the freebie land)   

Because of this I have committed to getting Poser in two days.  (2 days because while I doubt I will win it, it would be just my luck to win it two days after buying it lol). I will likely eventually do nearly everything in Blender, but I need to learn as much as I can about Daz and Poser both... and want to pass the torch on to those like CatHarr1.  I don't know everything but I don't want others to think that things which Daz makes annoying are impossible.  They are merely annoying, but do-able.  I want people to know they can be smarter than the marketting scenario and "free" really can mean "free". Knowledge should be free, I think.   Meanwhile I will throw my real support on the business side of things to folks like Hivewire and Fantasy Attic who treat the developer and the community with a collaborative respect. 

I appreciate the feedback Hedd!  I love your directness.  Sometimes I wonder if I exactly understand the subtlety of the few words you use.  Lemme know if I have gotten it wrong sweet man!