AI Generated Images

Prompt: illustration, arcology made from trees, megastructure, metropolis in the background, colorful, brutalist architecture --ar 2:5

Still a bit in my green brutalism phase. I wanted to make a vertical design, something that really stands out. Also I really like the colors :)

Bing/dalle3, prompt was "megacity of hobbit high-rise structures, rooftop gardens, organic, waterways, deep depth of field, photo."

Definitely not what I was going for but really cool.

Here’s my submission:

The prompt was

nature filled city with biomimicry architecture in a 512k mac game

Generated with bing image creator.

My first prompt was "floating cities with agriculture"
"floating cities" => it was too dystopian. Very artificial squared islands with tons of buildings. So I reversed it to "towns on floating islands" to enhance natural looking islands.
Without "palms" and "solar pannels" => it was like floating getthos. I added "palms" and "solar pannels" to make it paradisiac and futuristic.
Resulting prompt (Dall-e3) : "sea covered with towns on floating islands, palms, solar pannels, agriculture"

Prompt: Plant-based architecture, metropolis rising from an enchanted forest, clouds, grown, futuristic

Workflow/Prompt

I asked chat gpt + dalle

the topic is “Sustainable Ecumenopolis”. Come up with a novel concept for this.

The make some award winning high-quality images about it

then I rejected the first batch of result and said

vertical farming and solar panels are stale and boring.

the images should be award winning high quality and atmospheric.

then it generated some images. The one I selected for the competition was this

Render: Atmospheric night plaza in Biosphere Nexus. Architectural structures inspired by biomimicry surround the area. Trees with bioluminescent leaves light up the ambiance, with diverse citizens enjoying the scene.

If I win, the next topic will be "Torment Nexus"

Video

I had to scale the gif down with an online Tool which put me their logo in the bottom right corner.

Its ok If the use of tools outside of an classic AI Generator violates the rules of the contest and disqualifies me. I still had much fun trying it out!

The idea was to "fly" from outer space deeper and deeper into an Ecumenopolis (had to google thatone) until you "arrive" at the final destination deep into the Ecumenopolis, a grass field.

• AI Tool: Comfyui

• Model: dreamshaper_5BakedVAE

• Sampler: euler

• Scheduler: normal

• CFG: 7.0

• Samples: 40

• Positive Prompt: Its complicated.. See below

• Negative Prompts: bad anatomy, bad proportions, blurry, cloned face, deformed, frame, border, black rectangle, disfigured, duplicate, extra arms, extra fingers, extra limbs, extra legs, fused fingers, gross proportions, long neck, malformed limbs, missing arms, missing legs, mutated hands, mutation, mutilated, morbid, out of frame, poorly drawn hands, poorly drawn face, too many fingers, ugly, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, out of frame, ugly, extra limbs, bad anatomy, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck

• Negative Embeddings: FastNegativeV2, BAdDream

The Workflow is also embedded in this image:

The Workflow uses custom nodes which have to be installed to work

This is quite an interesting Workflow:

1. Base Image

I´ve created a Base Image with following Prompt:

documentary photography photo of closeup of a single grass blade, futuristic, grenn , ecological, hippie city background, natural lighting, from below/low angle, ARRI ALEXA 65, Kodak Vision3 IMAX

2. Send the created image back into the workflow

Via the image Sender and image receiver node of the Comfyui custom Module "Impact pack" you are able to send the generated image back into the workflow. With these nodes you are able to do a Generation loop which makes it possible to generate something like an animation by the creative use of inpainting.

3. Scale the genrated image down an pad it for outpainting

The previously generated image isscaled down to 50%. then we pad the image to get it back to its original size for the outpainting. combined with an outpainting controlnet you get a new image which "zooms" a bit out:

4. Now starts the Fun

You have in Comfyui an optional Setting where you can set the Image generation to an endless loop. It generates image after image until you say stop. As we have generated a feedback loop in our workflow we get the nice zoom effect for every following image.

And here is the point from above with the prompt "It´s complicated"

You can now modify the promp while the generation process is running and alter slowly the prompt. If you are a fast typer you can do this while the process is running or you pause, modify the prompt and start the process again.

My goal was to have a "Zoom in from Space" into the Ecumenopolis.

As the process works in the way that you zoom out from the first image, you have to think in "reverse" and modify the prompt gradually during the generation process so that you come frome the detail view of grass to the outside view of the Ecumenopolis from Space.

Surprisingly you only need a few iterations to get a nice effect:

4. Make a Video

The last step is to Generate the Animation with the Zoom Video Composer . This will generate you a .mp4 out of your single images. You also have many parameters to play with and to get different effects.

Solar-punk architecture, cocoon buildings, cityscape, city, rooftop view, greenery, glass, pedestrian, flower, urban, pods, resort, curves, unique, original, organic, perfect composition, concept art, sidelighting, depth of field, sharp focus, hard light, split shadow, vivid, brilliant, cinematic, dramatic, caustics, realistic, photorealistic, photo (medium) Negative prompt: solar panels, 3d, render, fake, illustration, painting, cartoon, drawing, dark, noir, desaturated, blurry, illustration, fake, plastic, toy, bland, smudge, cartoon, painting, drawing, fat, ugly, forehead mark, 3d, render, smooth, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name Steps: 30, Sampler: DPM++ 2M SDE Karras, CFG scale: 7, Seed: 1493947496, Size: 1280x768, Model hash: 70229e1d56, Model: juggernautXL_version5, Clip skip: 2, RNG: CPU, Version: v1.6.0

Used the bing thing again (I bonged the bing out of laziness?)

Prompt was: "Large clean mega city glass tower covering all, built on the ruins of the past, landscape style photograph. Night time"

Took 13 attempts, adding and removing words.