Leonardo AI, a popular generative imaging service, recently announced the launch of its new model, Leonardo Phoenix. This release comes shortly after Stability AI’s release of its SD3 model, which offered impressive adherence to prompts and text generation capabilities, but was marred by controversy surrounding its licensing terms and a misunderstanding of the human anatomy.
The creators of Leonardo were, as expected, enthusiastic about their latest release.
“We don’t want to overstate this, but we truly believe that Phoenix is one of the leading models in terms of its ability to faithfully follow a user’s input,” they wrote in an email announcement. The AI model update came with a revamp from Leonardo’s website.
Leonardo Phoenix is touted as offering better fast adhesion, clear and precise text within images, superior image quality and greater creative control. The update also introduces new features like “Improve Suggestions” and “Edit with AI,” which help users automatically convert basic suggestions into more detailed ones and instruct AI to change suggestions in specific ways, respectively.
However, due to the new architecture, some previously available features such as image guidance, elements, and photorealism are not yet supported. It is also important to note that token costs are temporarily lower for the new model, but will return to standard prices next Monday.
Ideogram Recognitions
While Stable Diffusion models require users to have the technical knowledge to install them on their local computers, Leonardo AI is easily accessible through its own website. This ease of use invites a natural comparison with Ideogram, MidJourney or Dall-E 3, which are also web-based AI image generators.
Before the release of Leonardo Phoenix, Decrypt compared these models—locally and using the web version—and concluded that “Ideogram may be the best image generator currently on the market.” When SD3 came out, our quick test concluded that it was “better than SDXL for many use cases, but not enough to replace it.”
Ideogram held strong in our subjective tests, but will Leonardo be good enough to beat it?
Leonardo versus Ideogram
To compare the models, we provide the same base prompt to both. We also activated a function available in both services to improve our prompt and generate better images.
Prompt interpretation
PromptBase: A man and a woman having dinner in a futuristic restaurant, illustration in the style of Vincent Van Gogh. The restaurant has a sign that says “Welcome to Emerge, by Decrypt.”
We wanted to test how well the models identify relevant artists and art movements, in this case, the artistic style of Vincent Van Gogh, who has a distinctive and easily recognizable technique. Leonardo surpasses Ideogram in terms of style. Additionally, its text generation capabilities were more powerful in this example.
Leonardo’s focus on style led to the fusion of the restaurant with the grass on a picnic.
Ideogram opted for a safer approach, focusing on scene rather than style. It is easier to correct Leonardo’s generation with inpaint. Ideogram would require more sophisticated techniques (such as using IPAdapter or Img2Img) to achieve acceptable results because the image does not resemble the style at all.
Winner: Leonardo
Prompt adherence and spatial awareness
PromptBase: A dog standing on a cat. On the left, there is a robot with a screen that says “Emerge”, on the right there is a horror-themed doctor with a syringe wearing a gas mask. The background mixes elements of emerging technologies such as DNA and genetics, binary code, space and alien exploration, electronics and futuristic technology.
Here we wanted to test how well the models understand the position of specific elements in the cue. They were both precise in their generations, understanding the position of each requested element, the background and the text.
Ideogram generated an image that combines all the elements and is more aesthetically pleasing. Leonardo generated a more accurate representation, which is evident in the small details. For example, Ideogram failed to generate good hands for the doctor, and the elements on the table are nothing more than messy abstract generations that look good but mean nothing.
Winner: Tie. Leonardo wins in terms of precision, Ideogram wins in terms of aesthetics.
Bonus round: We decided to start with this generation to test Leonardo’s “Edit with AI” feature. We asked Leonardo to “give the image a vintage ’70s style without changing its elements,” and he generated it almost perfectly on the first try, just missing the cat.
Importantly, this feature greatly modified the build and is not simply comparable to using IPAdapters in Stable Diffusion.
Photorealism
PromptBase: professional photo, close-up of a Caucasian man, wearing a black sweater, serious face, dramatic lighting, nature, gloomy, cloudy weather, bokeh.
Ideogram’s image was more photorealistic than Leonardo’s. He focused on the subject and all its characteristics, while Leonardo, again, focused more on style and other elements surrounding the subject such as shadowy aesthetics and dramatic lighting.
Winner: Ideogram
Feature Comparison
Both Ideogram and Leonardo are websites that allow the public to access their respective generative AI models. They also offer payment levels that enhance your capabilities. While the image comparisons are compelling, it’s also worth examining the technical details. After all, the deciding factor could go beyond the models themselves, encompassing each one’s overall service offerings.
Quick enhancement and image handling
- Prompt improvement: Both Leonardo and Ideogram offer prompt enhancement features to refine and improve user prompts. In Leonardo, this is called “Enhance Prompt”, while Ideogram refers to it as “Magic Prompt”. These tools aim to ensure optimal results by automatically expanding basic prompts into more detailed and effective versions.
- Image size: Leonardo AI supports larger image resolutions compared to Ideogram. For a 16:9 aspect ratio, Leonardo can output images up to 1376×768 pixels, while Ideogram is limited to 1280×720 pixels. This difference in maximum image size can be significant for users who require higher resolution outputs.
- Image Relationships: Both services offer a wide range of image aspect ratios to choose from. However, Leonardo AI offers an additional layer of convenience by providing references to aspect ratios commonly used on various social media platforms.
- Generation Speed and Quantity: While Ideogram’s image generation process tends to be slower, its free tier allows for more image generations. On the other hand, Leonardo AI’s free tier imposes stricter limits on the number of images due to the higher computational demands of its models. This trade-off between speed and quantity may influence your choice.
Advanced editing and customization
- Edit with AI: Leonardo AI offers an “Edit with AI” feature, which Ideogram currently does not have. This tool allows users to provide natural language instructions to modify their prompts in specific ways, such as “change the coat to red,” “add a blue hat,” “give it a vintage look,” or “add dramatic lighting.” The AI adjusts the indication accordingly, allowing for more precise control over the images generated. It is similar to the instructPix2Pix plugin that Stable Diffusion users have been using since last year.
- Background Expungement: Leonardo AI offers a background removal functionality, which Ideogram does not have. This feature can be particularly useful for users working with other image editing software or those looking to isolate and highlight specific subjects within their generated images.
- Image Enlargement: Leonardo AI provides an upscaling feature that allows users to enhance the resolution of their generated images. This capability is not currently available on the Ideogram platform, requiring the use of additional tools.
- Instant Image Generation: Leonardo AI offers an instant image generator similar to Meta’s “Imagine” feature or Krea’s instant generations. This tool generates images in real time as the user types their prompt, providing a more dynamic and interactive experience compared to traditional batch generation.
- Real-time animation and canvas: Leonardo AI allows users to create basic animations from their generated images and offers a real-time canvas for intensive inpainting and outpainting tasks. These features are not currently available on the Ideogram platform.
- Model training: Leonardo AI allows users to train their own models based on a curated selection of images. This feature requires significant computational resources and is not offered by Ideogram. This could be a significant advantage for users with specialized needs.
- Flexibility: Leonardo AI boasts a wide variety of models, with the Phoenix being its most capable offering, comparable to Ideogram’s proprietary model. However, the company also offers other proprietary models such as Leonardo Kino, Leonardo Vision, and a Lightning Model, as well as community-tuned open source models such as Dreamshaper and Absolute Reality.
Variety of prices and models
- Price per Generation: As for pricing, Ideogram’s paid version offers unlimited slow builds starting at $20 per month, while Leonardo’s Artisan Unlimited tier, which offers a comparable offering, is priced at $30 per month. For users requiring faster build speeds or additional features, Leonardo AI may be the most cost-effective option, despite its generally higher pricing structure.
Conclusion
Leonardo provides a complete user interface with a wide range of features designed for users looking to extract maximum value from their generated images. The integration of natural language processing for rapid editing and generation manipulation is also a significant advantage. For advanced users, especially those willing to invest in premium services, Leonardo is clearly the most compelling option, especially given its extensive feature set.
However, for casual users primarily focused on generating visually stunning images with minimal effort, Ideogram’s simplified approach, along with its AI-enhanced suggestion generation capabilities, may be a better fit.
While Leonardo appears to show a slight advantage in generating longer text strings within images, the difference is not substantial.
Ultimately, the decision between the two services may depend on individual priorities, with Leonardo being best for those looking for comprehensive creative control and advanced tools, while Ideogram appeals to users who prioritize a streamlined but simple.
In our opinion, Leonardo really takes the crown from Ideogram. While Ideogram can sometimes produce superior images, the number of features Leonardo provides is compelling. Whether you are new to generative imaging or a professional, Leonardo can meet your expectations.
Overall Winner: Leonardo
Edited by Ryan Ozawa.
