AI Typography Generation, Leonardo AI & Consistent Character GPT Updates
Turn AI into your ultimate creative collaborator
Midjourney V6 vs DallE 3 Typography Generation Comparison
Typography or text generation in an image has been impossible, until now.
Sure Ideogram.ai has been fun to play around with but we need a more serious tool.
DallE 3 was one step ahead at this and now Midjourney V6 has joined the party. So I ran some experiments
Here is the verdict:
1. Both are equally accurate at creating short texts
2. Midjourney V6 is superior in aesthectis (obviously)
3. Dall.E 3 is best at prompt adherence.
Example prompt:
๐พ๐๐ฃ๐๐ข๐๐ฉ๐๐ ๐จ๐๐ค๐ฉ of ๐ผ ๐ข๐๐ก๐ ๐ฌ๐๐ฉ๐ ๐ ๐ง๐ช๐๐๐๐ ๐๐๐๐ง๐ ๐๐ฃ๐ ๐๐ฃ๐ฉ๐๐ฃ๐จ๐ ๐๐ฎ๐๐จ ๐ฌ๐๐๐ง๐๐ฃ๐ ๐ ๐ฌ๐ค๐ง๐ฃ ๐ก๐๐๐ฉ๐๐๐ง ๐๐๐๐ ๐๐ฉ ๐๐ฃ๐ ๐๐๐๐ฃ๐จ, ๐จ๐๐ฉ๐ฉ๐๐ฃ๐ ๐ค๐ฃ ๐ ๐๐ง๐๐๐๐๐ฉ๐-๐๐ค๐ซ๐๐ง๐๐ ๐๐๐ฃ๐๐ ๐๐ฃ ๐ ๐๐๐ข๐ก๐ฎ ๐ก๐๐ฉ, ๐๐๐๐ฃ๐๐ค๐ฃ๐๐ ๐จ๐ช๐๐ฌ๐๐ฎ ๐จ๐ฉ๐๐ฉ๐๐ค๐ฃ. ๐ผ๐ฃ ๐ค๐ก๐ ๐ฃ๐๐ค๐ฃ ๐จ๐๐๐ฃ ๐ง๐๐๐๐จ "๐๐๐๐๐๐๐๐" ๐๐๐ค๐ซ๐ ๐ฉ๐๐ ๐ฉ๐ง๐๐๐ . ๐๐๐ ๐ข๐ค๐ค๐ ๐๐จ ๐๐๐ง๐๐ ๐๐ฃ๐ ๐ฃ๐ค๐จ๐ฉ๐๐ก๐๐๐, ๐ฌ๐๐ฉ๐ ๐๐๐๐ค๐๐จ ๐ค๐ ๐ ๐ฉ๐ง๐๐๐ฃ ๐๐ฃ ๐ฉ๐๐ ๐๐๐จ๐ฉ๐๐ฃ๐๐.
Try it out and share your results with us on LinkedIn, tag @sachinkamathai
Example prompt:
Cinematic shot of an individual wrapped in a thick coat, standing in a snow-covered forest clearing at night, looking at a neon sign saying "SILENCE" hanging from a bare tree, with snow gently falling.
Example prompt:
Cinematic shot of a male with a mischievous grin and bright, curious eyes wearing a colorful, patchwork coat, amidst a deserted carnival at sunset. A flickering neon sign that says "DREAMS" hangs lopsidedly over a derelict carousel. The atmosphere is whimsical but melancholic, with long shadows stretching across the abandoned rides
AI Video Alternatives
Ever heard of Cascade Diffusion Models?
Well it does a pretty darn good job compared to Runway and Pika.
This is an open source model now available on Replicate for anyone to try and so I wanted to share what it can do.
I used the above generated image from Dall.E as a reference for this model.
Additionally I used Topaz AI to upscale and increase fps to 30 ( I never skip this step for any AI generated video)
Here is the result.
Pretty right?
Leonardo vs Runway - AI video battle
Leonardo AI which is powered with Stable Diffusion is emerging as a clear contender for AI motion, and this is just V1.
Note: For 1:1 comparison the director mode/motion brush on the Runway was not used but the author.
Credits @maxescu
Leonardo is evolving into the quintessential all-in-one platform for generative AI, offering a remarkable array of models and tools for diverse applications.
This platform encompasses a wide range of features including ControlNet (with capabilities like image-to-image conversion, depth perception, edge detection, pose estimation, and text-to-image transformation), consistent character generation, sketch-to-high-fidelity rendering, creative upscaling, a comprehensive API, video processing, real-time AI operations, an expansive infinite canvas editor, advanced texture generation, and even custom model training.
We invite you to explore the depths of Leonardo and discover the extensive possibilities it offers for creative and technical endeavors.
New Update to Consistent Character GPT๐ฅ
Now you can create more than 4 variations and also full body pose.
Just ask for 'Generate full body pose of [character name] wearing [type of shoe] and [action]'.
TRY IT HERE - it is completely FREE with ChatGPT plus!
Dropping my 'GTA - The Ultimate AI Race 2024' teaser trailer.
AI tool stack I used:
โก Midjourney V5.2, Stable Diffusion - Image generation and stylization
โก KreaAI/ Magnific AI - AI image enhancement
โก Stable Video Diffusion, Runway, Magic Animate, HeyGen - AI motion & animation
โก AI Video Faceswap
โก Topaz AI - AI Video upscale and enhance to 60fps
More to come. Stay tuned.
Happy Weekend!