Stability AI management and Stable Diffusion 3.5 vs Midjourney

Stability AI manages capital from a $101 million raise while navigating competition with Midjourney. While Midjourney offers superior lighting and composition, Stable Diffusion 3.5 provides unmatched customization through tools like ControlNet and LoRA.

Stability AI management and Stable Diffusion 3.5 vs Midjourney

Stability AI faces pressure to turn its massive user base into profit. The company reported more than ten million daily active users following the release of Stable Diffusion. However, only a small percentage of these users actually make money for the company. Most revenue comes from the API and the Dreamstudio web interface. Mostaque says dozens of companies ask for foundation models as a service every day. He believes the open model trajectory will be clear later this year. Mostaque also claims that nine large hedge funds contacted him about collaboration in a 48 hour period. This comes as the company manages its capital after the $101 million raised in December 2022. The startup burned through a significant amount of that money to acquire Clipdrop and provide server time for researchers. Two investors now doubt a new funding round that would value the company at four billion dollars. Mostaque wants to remain CEO and envisions an IPO in the next few years. He thinks the open model approach is superior to closed systems. The company has 26 open positions, which is fewer than the 36 positions at OpenAI. The company must manage its expenses carefully because it already spent a significant portion of the $101 million in capital that it raised from various venture capitalists in late 2022.

Fidelity and model lineage

Midjourney produces images with better lighting and composition than Stable Diffusion. The tool works through Discord and requires a subscription. Stable Diffusion 3.5 is the latest generation of the open-source models. The models follow several lineages: SD1.5, SDXL, Flux, and SD3.5. SD1.5 is lightweight. SDXL provides a balance of quality and versatility. Flux requires more hardware power. Stable Diffusion 3.5 remains the latest open-source option for users who want control. It belongs to a lineage that includes SD1.5, SDXL, and Flux. While SD1.5 is lightweight, Flux requires more power. The tool allows for custom training that Midjourney can never match. However, the results from generating prompts often include distorted hands, teeth, or limbs. Most users find the Midjourney algorithm produces more realistic results than the base Stable Diffusion models. You can use SDXL to get better photorealistic images with models like Juggernaut XL. Midjourney users often use the –no command to remove unwanted elements from a scene. Stable Diffusion requires more skills to maximize the potential of the tool. If you want to use SD3.5 or Flux seriously, you need a GPU with 12GB of VRAM or more. Can Stable Diffusion eventually match the aesthetic quality of Midjourney?

Cost and customization

Midjourney pricing ranges from $10 to $120 per month. Stable Diffusion APIs start at $9, while higher plans cost $49 or $149. You can also run Stable Diffusion locally for free if you have the right hardware. An RTX 3060 with 12GB of VRAM costs between 30,000 and 40,000 yen. An RTX 4060 with 16GB of VRAM costs between 60,000 and 80,000 yen. Using a GPU for two hours a day adds 200 to 400 yen in electricity costs per month. You can also use cloud options like Mage.space for as little as 450 yen per month.

Feature Midjourney Stable Diffusion
Monthly Price $10 to $120 $9 to $149
Hardware Discord Account 4GB to 12GB+ VRAM
Learning Curve Easy Steep
Customization Moderate 10/10

For those who want maximum control, Stable Diffusion is the better choice. It allows for unlimited customization through ControlNet and LoRA. Midjourney provides an easier experience but less control over specific details. Most users find the $30 plan for Stable Diffusion most useful for their needs. I recommend starting with a cloud service like Mage.space to test the technology before buying a GPU. Testing these tools on a budget prevents wasted capital.

airtrain.ai
airtrain.ai

The airtrain.ai newsroom covers AI research, models and the tools built on them.

More on this topic

Stay ahead of AI

Get the week's most important AI stories delivered to your inbox every Monday.

No spam. Unsubscribe anytime.

More Stories