TL;DR:
- Auraflow, a new open-source AI image generator, aims to outperform Stable Diffusion 3 (SD3) with its Apache 2.0 license.
- Comparison: Auraflow excels in impressionistic and fantastical styles, while SD3 is better at hyper-realistic and dynamic scenes.
- Hardware Requirements: Auraflow needs more VRAM (up to 35 GB) compared to SD3 (6 GB), making SD3 more accessible.
In the rapidly evolving world of AI image generation, a new contender has emerged: Auraflow. Developed by the generative media company FAL AI, Auraflow is gaining traction with its open-source Apache 2.0 license. This article pits Auraflow against Stability AI’s Stable Diffusion 3 (SD3) to see which model reigns supreme.
The Rise of Auraflow
Auraflow, released last week, is making waves in the AI community. Its open-source nature allows developers to tweak, modify, train, and even profit from their work without licensing fees. This freedom is crucial for speeding up development cycles in competitive industries.
Quote from FAL AI:
“We are excited to present you [with] the first release of our Auraflow model series, the largest yet completely open-sourced flow-based generation model capable of text-to-image generation.”
Training and Performance
Auraflow underwent rigorous training over four weeks, including pretraining with images of various sizes and resolutions. The result? A GenEval score of 0.64, boosted to 0.703 using a prompt-enhancement pipeline similar to DALL-E 3. Despite its impressive performance, Auraflow is still in beta (version 0.1).
Hardware Requirements
Auraflow requires a beefy GPU with around 12 GB of VRAM to run its fp16 version, compared to SD3, which runs fine on just 6 GB VRAM. However, FAL AI is working on a more manageable model.
Quote from FAL AI:
“Smaller models or MoE’s might be more efficient for consumer GPU cards, which have a limited amount of compute power, so follow closely for a mini version of [this] model that is still as powerful yet much much faster to run.”
Art Styles and Creativity Prompt:
“A detailed painting of a sunset over a tranquil lake, the sky filled with hues of orange, pink, and purple, a wooden pier extending into the water, a person sitting at the end of the pier with a fishing rod, surrounded by tall grasses and wildflowers, the overall style is impressionistic with bold brushstrokes and vibrant colors.”
- Auraflow: Captures the impressionistic style well but lacks detail in the person and nature.
- SD3 Medium: High attention to detail but less pronounced impressionistic style.
- Winner: Tie. Auraflow follows the impressionistic style more closely, but SD3 is more detailed.
Realism Prompt:
“A high-resolution photograph of a bustling city street at night, neon signs illuminating the scene, people walking along the sidewalks, cars driving by, a street vendor selling hot dogs, reflections of lights on wet pavement, the overall style is hyper-realistic with attention to detail and lighting, a neon sign says ‘Decrypt.’”
- Auraflow: Captures the vibrant nightlife but lacks sharp details.
- SD3 Medium: Provides high detail and clarity, making it the better model for this prompt.
Illustration Prompt:
“Hand-drawn illustration of a giant spider chasing a woman in the jungle, extremely scary, anguish, dark and creepy scenery, horror, hints of analog photography influence, sketch.”
- Auraflow: Creates a dark and creepy atmosphere but lacks detail.
- SD3 Medium: Offers a highly detailed and scary portrayal, making it the better model for this prompt.
Prompt Adherence Prompt:
“A surreal digital artwork of a floating island in the sky, the island covered in lush greenery and waterfalls cascading into the clouds below, a small castle at the center of the island, bridges made of light connecting to other floating islands, the sky is filled with colorful hot air balloons and mythical creatures, the overall style is fantastical with dreamy elements and glowing effects.”
- Auraflow: Captures all elements well but lacks detail in some.
- SD3 Medium: Highly detailed but weaker prompt adherence.
- Winner: Auraflow captured all the elements in the prompt.
Spatial Awareness Prompt:
“A dog standing on top of a TV showing the word ‘Decrypt’ on the screen. On the left there is a a woman in a business suit holding a coin, on the right there is a robot standing on top of a first aid box. The overall scenery is surreal.”
- Auraflow: Creates a surreal scene but with less refined details.
- SD3 Medium: Provides a highly detailed depiction but less imaginative.
- Winner: Tie. Both models provide all elements of the generation.
Anime and Manga Prompt:
“A female ninja fighting against a strong samurai in ancient Japan, anime, manga, highly detailed, colorful, dynamic.”
- Auraflow: Captures the dynamic and colorful elements but lacks adherence.
- SD3 Medium: Provides a more detailed and dynamic depiction but fails to capture the scenery.
- Winner: SD3 Medium provides a more detailed and dynamic depiction.
Conclusion: The Future of AI Image Generation
Auraflow excels in capturing impressionistic, fantastical, and whimsical styles, while SD3 Medium is better at providing detailed, hyper-realistic, and dynamic scenes. Auraflow’s Apache 2.0 open-source license makes it attractive for fine-tuners, but it requires more VRAM than SD3.
Quote from FAL AI:
“Some even boldly announced that open-source AI is dead. Not so fast!”
Comment and Share:
Which AI image generator do you think will dominate the market in the future? Share your thoughts and experiences with AI and AGI technologies in the comments below. Don’t forget to subscribe for updates on AI and AGI developments.
You may also like:
