Amazing FREE AI Image Generator: FLUX.1 (Can it challenge Midjourney?)

The video introduces Flux, a new open-source generative AI model developed by Black Forest Labs. The company was founded by former members of the Stable Diffusion team, and Flux is their first AI model. It is positioned as a strong competitor to MidJourney, offering powerful text-to-image generation capabilities. The narrator showcases various examples of images generated by Flux, including a street scene in Freiburg, a Victorian tea party with spiders, and a film-like photo of a man's eye. Flux has been benchmarked against other AI models, and according to the Elo score, it outperforms its competitors, including MidJourney. The video also mentions Flux's three versions: Flux Pro, Flux Chanel, and a free model, with Flux Pro being a paid option that offers commercial use.

The narrator begins a structured comparison between Flux and MidJourney, focusing on natural language understanding. They test the prompt 'photo of a horse is riding the man' using both Flux Pro and Flux Chanel models. Despite adjustments, the Flux Pro model struggles to accurately represent the prompt, while Flux Chanel produces fast but less impressive results. MidJourney, however, does a better job of adhering to the prompt, albeit with some unexpected retro color choices. The narrator concludes that for this prompt, MidJourney is the winner due to its closer alignment with the intended output.

In the second comparison, the prompt 'an angry woman is chasing a dog' is tested. Flux models perform exceptionally well, accurately depicting the woman slightly behind the dog, both characters showing the intended emotions. MidJourney, on the other hand, produces an image where both the dog and woman appear angry, with the woman slightly ahead of the dog, which deviates from the prompt. As a result, Flux models are declared the winners in this round for their accurate representation of the action scene.

The third prompt involves creating a cinematic photo of two women sitting in a cafe, one with red hair and leather clothes, the other in a suit. All three models—Flux Pro, Flux Chanel, and MidJourney—perform well, but MidJourney stands out due to its cinematic color grading and precise representation of the characters' attire. The narrator praises MidJourney for its superior prompt understanding and image quality, making it the clear winner in this round.

For the prompt 'upside down Egyptian pyramid,' the results vary significantly between the models. Flux Pro impresses with its abstract interpretation, closely matching the narrator's vision despite some scale issues. MidJourney delivers a more literal, yet visually compelling result, while Flux Chanel falls short of expectations. The narrator favors Flux Pro for its creative approach, naming it the winner of this challenge.

In the fashion photo challenge, the prompt involves a man wearing a t-shirt with blue and purple polka dots and a brown hoodie. Flux Pro delivers a nearly perfect image, with the prompt fully understood and accurately represented. Flux Chanel veers into a more cartoonish and 3D direction, which is not what was desired. MidJourney's output is acceptable but flawed, particularly with the hoodie. Therefore, Flux Pro is declared the clear winner for its precise prompt understanding and execution.

The narrator tests the models with a prompt featuring 'John Hamm in a Batman suit sitting in his Batcave with his mask on the ground.' MidJourney delivers the best likeness to John Hamm, despite some imperfections. Flux Pro fails to capture Hamm's likeness and does not place the mask on the ground as instructed. Due to its closer alignment with the prompt, MidJourney is declared the winner for this challenge.

The photo realism test sets the guidance value to two in Flux Pro, balancing image quality with prompt understanding. While Flux Pro performs well, MidJourney still comes out on top, despite some realism issues with textures and skin. Flux Chanel reverts to a more cartoonish style, making it less competitive. The narrator concludes that MidJourney, even with its flaws, remains the best in this round for photo realism.

The prompt 'photo shot from above, hands are playing piano' tests the models' accuracy in rendering details. All three models—Flux Pro, Flux Chanel, and MidJourney—perform admirably, with minor flaws in some fingers' visibility. The narrator notes that despite these small issues, all models successfully capture the prompt's essence, with no clear winner in this round.

The next challenge involves a cinematic photo of an Indian woman holding a bow and arrow, ready to shoot. MidJourney impresses with its detailed and epic depiction, making it the clear winner. Flux Pro also delivers a strong result, but with some blurring issues that reduce detail clarity. Flux Chanel's result is less impressive, with significant flaws compared to the other models.

The sports photography prompt features a female volleyball team jumping to hit the ball. Flux Pro performs exceptionally well, capturing the action with accurate anatomy and a diverse audience, despite some facial issues. MidJourney struggles with anatomy and positioning, leading to less coherent results. Flux Chanel, while not bad, lags behind the other models. The narrator concludes that Flux Pro is the best in this sports photography challenge.

The final challenge tests text rendering with the prompt 'Jungle Fire as a brand on the bottle of a hot sauce.' All three models—Flux Pro, Flux Chanel, and MidJourney—deliver excellent results, but Flux Chanel stands out with near-perfect text rendering and branding. MidJourney also performs well but has a slight error, and Flux Pro is strong but presents two bottles instead of one. The narrator gives the win to Flux Chanel for its exceptional text rendering.

The video concludes with a summary of the challenges, noting that Flux and MidJourney are evenly matched, with each model winning five rounds and one tie. Flux proves to be a strong contender, especially in natural language understanding and text rendering. The narrator emphasizes that MidJourney still holds its ground in photo realism and detailed prompt execution but will need to innovate to stay ahead as Flux continues to improve.