All artworks and photos are copyright LightBWK 2021

10 years ago I am as puzzled as most people when it came to applying color. If the future-me told the 10 years-ago-me that I will write an ebook about color, the past me will look at him with squinting eyes, puckered lips and a hand scratching my head.

[insert “say what?!” meme here]

Boy, have we learned color the wrong way.

One should not approach learning color by learning color theories. It is boring, demotivating and often hard to apply in your artwork. The correct approach is learning to see colors before anything else.

Then I stumbled upon a very important discovery. In those few hours I mastered color! That was one of the most profound Aha! moment in my life.

I made a course about it (now no longer available). This ebook is my second attempt to impart the knowledge to more people.

What if, I tell you that you can name all colors on this color picker. It is possible.

That is a lot of colors right? All the colors on the picker have unique meaning to me now. Reading my ebook, you will experience that as well.

New depth, more nuances.

Another very important skill to master is the ability to color pick from anything, then later recall the color accurately. Just like that one moment when I saw a nice blue-green color on the dress of a beautiful lady, I still can recall and color pick the same color digitally today. This skill will help you recall many things more vividly. You will experience the world the way I see it. That alone is priceless.

Go shopping for new clothing, you can name any color you see. Play video game, you know the color names on your favorite heroes. Saw a cool artwork on Twitter, you know the colors he used even before you use a color picker. You can see color channel numbers when you decide to go wild with your color picking. SICK!

So what does it take to master color? What makes color stirring one’s soul?

When it comes to your digital tools, I can here people say: “Nah, I know color picker. It is just a color picker.” Boy, how wrong they are. In order to program a color picker, the developer must know quite a bit on how color works IRL and how to transform it into the color picker (unless he just pop in a framework without prior knowledge). This is called color model. The color model will mandate how you interact with the color picker. Different color picker types will either give you advantages or will cripple your color selection as well.

When you color pick from an image, it feels easy with any tool. But when you color pick from a set of requirements or designing a color palette from scratch, you will be creating a new world with its own sets of rules. This is the hard part. Fortunately more than half of the ebook will tackle this problem. This is the part where almost every color learner will face difficulty. When we leverage our digital color tools, picking soul stirring colors become easy. Things like color harmony and color unity are the result of proper color picking.

There are many ways to build a color palette, yet the foundations are few. Since they are few, learning them should be very quick. But they are only worthy when applied. Mastering those foundations will strengthen your ability to create soul stirring artworks.

A bit about the ebook.

The ebook is beginner friendly and useful for advance color designer. The topic starts from easy to very advance. In every step to color mastery, you will do simple challenges and experience new insights from them. You will land into advance topic without knowing it. The ebook also links to many topics you can find on the web. Including a link to some funny color names. 

If you don’t like long winded color theories and prefer knowledge you can apply directly, this is the ebook for you.

Soul Stirring Digital Color Mastery (SSDCM) is a shortcut to master color in digital medium. An ebook the past me wanted to read badly. Now that I have written the ebook, I am handing you the magic key to unlock soul stirring colors in your life.

Take full control of your colors.

One of my editor even wrote:

Many people were involved in the making of this ebook. Thanks guys, this ebook is miles better with your contributions.

Get Soul Stirring Digital Color Mastery ebook now.

Let’s start with Ryu!

Freestyle Overview

Do you like fluffy?

Edge nodes

Toon Shader from Scratch

Pixel Art? (3D to 2D to pixel art)

Above: pixel art version

“The making”

Have fun licking your screen.

Remember what Lee, Daniel and I did back in the end of 2012? To refresh your memory, here is what happened.

• Freestyle wasn’t in Blender yet.
• Blender NPR just started.
• We made Edge nodes.

Yes, the pre-freestyle line art era. Edge nodes took us about one month to publish. In that one month I saw the full potential of learning pixel math. It was way too cool to put into words. Let’s see some images.

The table line art render above cannot be done properly with Freestyle. As someone who wrote a comprehensive Freestyle course, trust me, freestyle cannot produce better line art compared to pixel math. Pixel math’s quality is almost hand drawn. To get this render, everything is done in pixel math via the compositor.

The power of pixel math isn’t limited only to the compositor. You can use it in material node as well.

A popular understanding is that you need to plug images into both inputs of mix node. Yet you don’t have to. You can drive the color mixing using only grey value inserted into factor input.

Yes, yes, I can hear some voices saying: “Nice render sir, but mix node is simple.”

No. Mix node is not simple. It is not too hard either. Mix node is just not explored and exploit to its full potential. Plus a lot of artist only understand few blend types, they are:

• Mix
• Add
• Multiply
• Overlay
• Screen

But what about the awesome Burn and Dodge blend modes? Do you know that the Soft Light blend mode is way better than Overlay blend mode? Do you know that Overlay has Multiply algorithm in it?

We are traversing too deep too soon, let’s back up a little.

Let’s go back to Mix node. The Mix node that we use now is in RGB. We actually can make Mix blend mode in HSV.

Do you know that mix blending algorithm is in almost all blend mode available in Blender? Only one blend algorithm is without the Mix algorithms which is Dodge blend mode. Another note, mix algorithm looks different in other blend modes. Thus almost everything is a remix of the Mix blend mode.

Other than the cult like node hacking, blend modes are very fun as well. For example we can use the Difference blend mode to solve spot the differences puzzle.

Note: Pixel Math is not limited to non-photorealistic application. It is heavily used in VFX & motion graphics.

As you might have noticed by now, pixel math can be applied extensively (Edge node below) or sparsely (material node example above).

Pixel Math (noun): Mathematical equations & algorithms which determine color behavior when two or more colors are layered on the same pixel location. Commonly known as blend modes.

If there is a way to learn pixel math, then it is first through the understanding of the basic blend algorithms provided via Mix node. After that, you can pick your own path. Solve some compositing problems. Invent new solutions to old problems. Take different point of view or a new paradigm to solve them. That’s where my Pixel Math e-book comes into the picture.

Up until now, blend modes are not well documented. The descriptions are often too vague. Without knowing the algorithms that produced them, we can’t say for certain that,

“this color blends that color, at that factor value, will resulting in that final color.”

My Pixel Math journey has been like this:

The e-book shows the top of the iceberg (describe the blend modes in easy to understand terms), opens your eyes to common blend algorithms (extracted directly from Blender source code) and invites you to make your own from what you have learned (the journey to the abyss).

I can say for certain that the people who have read the e-book will say this to you:

“You must be mad for not knowing pixel math deeper and earlier.
Almost everything you do in Blender involved pixel math.”

And just last week (week of June 14th 2015), 1 artist told me:

“Oh yeah, I think you’re touching on some shit I’ve been ranting about forever.”

After seeing the e-book, he was venting how come there were so little information about something so essential before the e-book. Yet he was not the first person telling me that last week. In fact he was the 3rd person.

As a digital artist, this is an important book. A Must Get.

You can get Pixel Math e-book via Blender NPR Store.

Camera Projection is all about finding perspective. In Blender we have tools like BLAM for camera projection. BLAM can be great for squarish objects, but when the scene has more organic shapes and forms, don’t blame BLAM for wasting your time. BLAM isn’t designed for all camera projection cases. Most of the time, manual camera projection is faster and more accurate.  Manual camera projection starts with better understanding of perspective, which is the fundamental of all visual art-forms. Let’s hop into the details.

What you need to know?

The following terms are just the basics, so there is a good chance that you already know them:

Horizon line or view height: The apparent line that divides the earth and the sky.

Perspective lines: The lines that point toward the horizon at different angles.

Vanishing point: The point where the angular perspective lines of an object visually continue past its edges and eventually converge on the horizon line.

Manual Camera Projection steps:

Setup the camera

• Reset camera position and rotation | (Alt G > Alt R),
• Translate camera up 5 unit in Z | (G > Z > 5),
• Add an empty at the (0,0,0) | (Shift A > Empty > Plane Axes),
• Parent camera to empty | (RMB Camera > RMB Empty > Ctrl P > Parent to Object),
• Adjust camera end clipping to 10000 (10km) | (RMB camera > Properties window > Camera Object Data > Clipping End > 10000).

This setup is to make us control the camera from the front and not from the origin point. This can also be done with 3D cursor as pivot, but it is harder to position 3D cursor in blank 3D space.

Setup viewport/3D View

• Make the grid floor big | (N > Display > Grid Floor > Lines > 1024)
• Change scale to 2 or more. Note: Horizon is approx 5km as seen from low altitude.

This process is done to enable us to align horizon of the image with the one in the viewport. You also need this to align the floor.

Setup the image

• Import image. | (N > Background Image > Open)
• Change axis to “Camera.”
• Image alpha, 0.5 is good.
• Image to Back (you may need to switch this front or back based on your view)
• Change aspect ratio to “fit” [extremely important, else you’ll waste your time building scene from a stretched image.]
• Adjust render size to image size. | (properties window > Render tab > Set X/Y size ) [Optional]

Approximate camera focal length

• Often times the default is a good start.

Tip: Real photos have a fix perspective without shift. To challenge yourself, rebuild an anime background with perspective shift and distortion.

Tip: For serious camera projection work, camera sensor size, lens type, focal length & shutter speed must be determined. Please refer to your camera manual for sensor size & lens type, or look up for camera reviews online. Focal length & shutter speed can be obtained from image EXIF.

Find the floor or the horizon (View height) or a wall

Horizon

• Select the empty,
• Rotate X to align horizon line,
• Increase height if needed.

You can sketch the position of the horizon on the image, but make sure you save it to another copy. Finding horizon or view height is easy. When you see lines converge to a flat horizon line, the view height is close. Another way is find the top side and the bottom side of the object in the image. View height is when both top and bottom start to disappear.

Note: The green line is image’s horizon. View Height (see Z) is approx from the vehicle height where the horizon line crosses it.

Floor

• Select the empty,
• Rotate X to align horizon line.
• Now rotate around the Z axis, make sure the grid on the image align with floor.
• If only 1 side is aligned, increase or decrease the height.

Often this method is used when you are not sure where the horizon is or you have a 3 vanishing point perspective situation. This is the best method when the image has some camera roll. A lot of times the floor isn’t a floor, it can be a beam frame, or a ceiling block that is parallel to the floor.

Note: This example has quite a bit of camera roll. Align the left edge first then adjust the rest.

Wall

This process is the same as the floor method, but your aim is to align the bottom corner of the wall to the grid.

• Get bottom edge aligned.
• Add a grid and rotate 90 degrees.
• Delete the face only. | (Edit mode > X > Only Faces)
• Align the grid to match image.
• Adjust empty’s height to add or remove wall skew.

Tip: The grid floor is your friend.

Tip: Properties sidebar > Transform, is great for controlled transform.

Tip: There are many other methods to align the camera (the image). Basically any geometry that has the approx shape of the object in the image will be useful.

Build the scene

When building objects inside the scene, you need to have 2 viewports. The 1st is for viewing from camera, and the second is free moving 3D viewport to be able to judge depth while modeling.

Tip: Camera projection is never perfect, always a just good enough process.

More Examples:

Note: The only perspective information that I have here is the view height, it can be seen on the rocks on the right bank.

Note: View Height is an approximation, we cannot get the precise data. Since POV is the highest point, the conifers after the dip must be lower than us. Note also the red lines, the mesh under it has the same width.

Note: Build the scene with another viewport, not many VFX software can have 2 synced viewports.

https://www.youtube.com/watch?v=hnSMXcwF6-g

Overall mesh view.

Note: Camera Projection project done by Amir for Mercy Malaysia.

Note: Dolly Zoom (also known as Vertigo Effect) with camera projection

Image Credits: 

Inside Domino Sugar Factory by Jason Eppink | CC-BY
Narrows nature by Jon Sullivan | Public Domain
Utah Sugar Company | Public Domain
Icefield Parkway Scenery – Rocky Mountains, Adam Jones, Ph.D. | CC-BY-SA
Waterfall by Peter Rivera Stamford, Connecticut| CC-BY
Jacob’s Ladder in Watkins Glen State Park By Beyond My Ken | GNU Free Documentation License
Beach ball by Juanedc from Zaragoza, España | CC-BY

Article Credit:

Planned & written by Light BWK with Amir‘s technical help. Extra editing by Lee Posey.

]]> https://studiollb.wordpress.com/2015/04/06/manual-camera-projection-basics/feed/ 12 wkbong Manual Camera Projection Perspective Setup Camera Grid Floor Setup Image Horizon Floor Method View height method Approx View Height method Waterfall Overall view of the mesh Camera View https://studiollb.wordpress.com/2015/03/01/blender-texture-paint-it-is-getting-there/ https://studiollb.wordpress.com/2015/03/01/blender-texture-paint-it-is-getting-there/#comments Sun, 01 Mar 2015 14:14:43 +0000 http://studiollb.wordpress.com/?p=1024 It has been over 2 years in planning and finally it is coming into shape. Texture painting is almost ready to handle big NPR productions. Here are some of my results using it.

Warning: Heavy images ahead. It will take some time to load.

Title: I’m going there (says the ladybug on the grass)

This is done about 6 months ago. At that time, texture painting was so laggy. I barely could paint this. It took so long to just block-out the composition.

Title: Color splat, wow, landscape

This tiny doodle to check if lag is gone. Yes, it is faster.

This is an early screen capture. I was very lazy to UV unwrap the meshes properly. Won’t do that again.

Title: Junction on the hill

Gave it another run. Not bad, but there is a problem. I can never get smooth strokes. After hours of finding the cause of the problem, the cause, Blender openGL cannot update that quickly. Some filter or stroke smoothing algorithm needed, but not until Blender openGL is upgraded.

I don’t like this painting method, although the style is very close to background done by Studio Ghibli (Gouache and poster colors). It just took too long to emulate real pigment.

The initial meshes look like this. They are a combination of planes and 3D objects.

I painted more than what you can see. As this is an earlier screen capture, the lighting for the clouds is wrong.

Title: Hidden tomb

When you paint in 2D plane, it is very fast. And I found a way to cover up the jaggy strokes.

It is really flat. Painting is done in 2k x 1k, slightly above HD.

Title: The summer when we catch grasshoppers

This is painted in 4k x 2k resolution. Rendered at HD. Painting at that size is OK even in GLSL.

Early screenshot of the viewport. They are transparent planes.

Conclusion as of March 1st 2015:

Texture painting is really getting enjoyable in Blender. It is quicker and more responsive. Even with its limited toolset, knowledge learned from texpaint can be applied in almost any painting application. Yet, there are a lot of things that I have not told you. There are few weird stuff in texpaint that exclusively in Blender. They can be such a headache.

One of them is the weird image I/O in Blender. It is easier to take an outside image into Blender, but very hard to get image created inside Blender to go outside (discussed, no action). Any texture painted in Blender has 3 names in 4 different locations (bug reported, rejected, UI team TODO). If you have a textured brush, the painted color is often brighter with random hue-saturation-value shift (bug reported, fixed). If painting in the image editor, alpha lock won’t work. Smoothing or smudging, alpha lock won’t work either. Mask only view-able in Multitexture material mode.

When you paint in 4k x 4k textures, it is extremely laggy. Something with that number causing my system to lag. If done in 4k x 3k it will be as smooth as 4k x 2k. The ability to paint smoothly at higher resolution is a must. I’m thinking of 10k x 5k (horizontal pan) or 5k x 15k (vertical pan). If not, I’ll just stick with 720p and resize the painting to 4k with some filter applied.

Update March 3rd 2015:

There are few requests for me to clarify texpaint I/O problem. I’ll list them 1 by 1 as much as I can remember from my experience.

1. When you save a blendfile, blender never say that your textures are unsaved.
2. You have to specify file path for each image and it is very easy to miss 1 or 2 or more textures.
3. Blender warns “invalid path” on header if you try to save textures painted without path but never says how/where the path should be set.
4. To set file path you have to open image editor then press F3, after only that you can use the “save all images.” Very counter intuitive.
5. The specify file path should be together with “save all images” button. Just specify the folder and all images will be saved there with the name listed in the paint slot.
6. You’ll lost your image(s) if you try to pack textures mid painting.

Other problems

7. you need to subdivide a mesh quite a bit  for it to not have wavy texture (texture stretching). In 1 case, my original meshes were less than 100 faces, to solve the wavy texture problem they were subdivided to 22k faces.
8. Not sure if soften and smudge should have alpha lock because they don’t work at all.
9. Alpha lock don’t work when painting in image editor.
10. Painting mask isn’t viewable in GLSL view mode, only viewable in multitexture view mode (without alpha).
11. Mask strength and mask value work almost the same, why there are 2? ie: Mask value 0.5 is similar to Mask value 1 with strength 0.5.
12. Generated image: UV grid and color grid have very limited size, they cannot cover 4k x 4k image. Everything not covered will be background color (in BI).