Using Robot to Draw an Image Generated From AI Midjourney

This project is created as a course assignment at the California State University, Long Beach; taught by Behnaz Farahi: DESN 551: Materials, Tools, and Techniques of Prototype.

Group Members:

Khoa Le, Michael Shatto, Pedro Mateo

When this project was assigned, the due date was going to be around Halloween, so we wanted to do a Halloween theme drawing.

Programs:

• Illustrator (Tracing the Image to 6 Colors)
• Photoshop (Separate the Image into layers, Edit the Image)
• Rhino & Grasshopper Plug In (Control and Robot Arm)

Supplies:

• Tombow Brush Pen
• Robotic Arm

The first step was to familiarize ourselves with the robot. We use a spiral image on the internet to practice working with the robot.

After the group familiarized ourselves with the robot and learned how to use it, we start going onto Midjourney to generate images for inspiration. We didn't know what kind of image we planned to draw, so we typed out general Halloween-related keywords like "nightmare"; "horror"; "Halloween"; etc.

These keywords was able to generate the images above. We liked these ghoul creatures with some red accents. It gives an ominous feeling that is suited for Halloween time.

Building from the previous images, we wanted explore a different prompt. We decided to use new keywords, hopping to get better pictures, like "ghouls figure" and "halloween".

We didn't like the results we were getting, so we went to the previous prompt and explored more variations of that prompt. Eventually, it led to these landscape images rather than more variations of the ghouls. They were good images, but we prefer the ghouls figure more.

This is the image we chose to use for the drawing. We like the depth of the image. There are multiple ghouls lurking, and the mist gives an ominous look. It's a good image to layer multiple colors.

With the first attempt, we noticed the robot drew the negative space of the image and not the positive space, so we had to go into Photoshop to invert the image. After that, the robot was able to draw the positive space.

After seeing what the image would look like as a drawing, we start separating the image into layers: foreground, background, and accents.

As we refined our drawings, we had to edit the images to bring out or hide away the details of the drawing. 

We started experimenting with using different colored pens on the layers. After all of these attempts, we learned to draw the images from the order of background, accent, and foreground. That way, the ink of the foreground layer would appear on top. This color combination of purple, blue, and magenta was interesting. However, it did not gives the ominous feeling we were getting from the image, so we decided to use different color pens. 

After changing the pen colors, we tested them out. There was no contrast between the fore and background. The grey and black pen were too similar.

To fix the contrast issue, our group changed the gap between the pen lines and the line weight. For the background (grey marker), we let the line distance at the default gap of 3mm, and we lift the pen up 0.25mm compare to the other two layers to give it a lighter look.

For the accent layer (red marker), we didn't like how many red lines were. It started become the focus of the drawing rather than the figures. So we increased the line gaps to 4mm to reduce the red lines.

For the foreground (black marker), we made the gap only 1mm. We want this layer to stand out the most, because it's the subject. We made the line gap close enough to make it almost like coloring so it could contrast the background lines better.

After finishing the drawing, we noticed the change in line distance caused less over lapping of the lines. Which allows the drawing to look more detailed and improves the contrast of each layer.

This is the final drawing of the image. After all of the fine-tuning of the different variables, this was the most desirable result. 

Surface Level:

Make sure the surface the robot is drawing on is leveled. If it is not, try your best to make it level. If it's not possible, make sure the to pay attention to the line weight and recognized the imbalanced regions and be adaptive. We noticed on the left side of the drawing tends to have thickest lines and have a lot of ink bleed due to the left side being more elevated. The surface we were working on was on a decline from left to right, it was about a 1mm difference.

Therefore, in our final drawing, we had to lower the pen height every ever so lightly to make sure the line weight is consistent throughout the entire drawing.