How to pick the best GenAI tool for images

In this episode, we sit down with Sachin Kamath↗? and unpack the world of AI-generated images. We explore his favorite tools, how he uses them creatively, and how his students are applying their skills to professional projects. We will also look at a tool that addresses one of the key problems when generating with AI: consistency and control on characters' looks and poses.

You can subscribe to and support XR AI spotlight right on Substack, with new issues appearing directly in your inbox, plus bonus issues only available to our supporters.

If you’d like to sponsor this newsletter, and get your name in front of an engaged audience of professionals and creatives, just contact me here on LinkedIn ?? Gabriele Romagnoli.

Interview with Sachin Kamath

What are your favorite AI image-generation tools right now?

Sachin Kamath: My favorite tools for AI image generation depend on the use case. For quick ideas that I want to visualize fast, I go to Ideogram ??. It's free, simple to use, and has solid prompt adherence. MidJourney is my all-time favorite for high-quality visuals, especially in the inspiration stages. Lastly, for more control over things like composition and poses, I use Leonardo AI. Each tool excels in its own area depending on whether I need speed, creativity, or precision.

Why do you prefer MidJourney for some cases and Leonardo AI for others?

Sachin Kamath: MidJourney is perfect when you’re looking to be inspired. It’s amazing for conceptual work, where you want to explore different styles and aesthetics without needing exact control. But if I need something very specific, like a person in an exact pose or composition, that’s where MidJourney falls short. Leonardo AI, with its stable diffusion extensions, gives me the kind of detailed control that MidJourney doesn’t. It really depends on whether I need creative freedom or strict accuracy.

What’s your take on why MidJourney hasn’t yet implemented pose control like Leonardo AI?

Sachin Kamath: MidJourney is designed to spark creativity rather than give users exact control. They’ve positioned themselves as a canvas for creative concepts, so they prioritize spontaneous creativity over precision. That’s why they haven’t focused on adding pose control yet. MidJourney gives you unexpected, wonderful ideas, but if you want to dictate every little detail, you need to step out of it and use something like Leonardo AI or other stable diffusion tools.

What makes Ideogram stand out compared to other AI tools?

Sachin Kamath: Ideogram is like the best of both worlds. It understands prompts as well as GPT models do and generates visuals that are surprisingly high-quality for a free tool. It’s easy to use, which is great for people just getting into AI image generation. It may not give you the polish of MidJourney or the control of Leonardo, but for quick, initial ideas, Ideogram is a great place to start.

How do you use DALL·E in your creative workflow, and where does it fit?

Sachin Kamath: DALL·E is great for generating quick visuals, especially when I don’t need photo-realism or refined aesthetics. It’s useful for concept visuals during the early stages of creative briefings, like in presentations. However, it’s not perfect for high-quality outputs like MidJourney or Leonardo AI. DALL·E listens to your prompts well, thanks to its integration with language models, but the images it produces can be flawed and lack the visual polish of other tools.

What’s your experience with ComfyUI, and who should use it?

Sachin Kamath: ComfyUI is an incredible node-based tool, but it’s more for those who want granular control over their image generation process. It’s especially powerful if you know exactly what inputs and outputs you’re looking for, and you want to build intricate workflows using stable diffusion extensions. But it’s not for everyone—if you’re new to AI tools, it can be overwhelming. Still, if you’re looking to experiment, platforms like OpenArt AI ?? have beginner tutorials.

Is it necessary to dive into complex tools like ComfyUI from the start?

Sachin Kamath: Not really. You can achieve a lot with web-based applications and existing open-source models on platforms like Hugging Face or Replicate??. Many use cases don’t require you to install ComfyUI locally or dive deep into node-based workflows. It’s really about knowing what you want to achieve. If you just need quick visuals or creative iterations, tools like MidJourney or Leonardo will suffice without all the complexity.

How do you tackle the issue of consistency when generating images with AI?

Sachin Kamath: I’m obsessed with consistency in AI-generated images. Consistent style and character are key when telling a story or building a storyboard. Before MidJourney introduced character reference parameters, there was no easy way to achieve consistency. That’s why I built the Consistent Character GPT, which became a top tool on the GPT Store. It’s great for creating animation-style characters, and for photo-realistic work, MidJourney’s latest updates are getting better, though there are still some variations.

How do the tools help, and how much is up to the creator for consistency?

Sachin Kamath: When it comes to character consistency, it’s mostly about using the right tools. There’s little skill involved—it’s all in how you leverage the technology. Tools like MidJourney’s updated character reference feature or face-swapping techniques give you the control needed for consistency. It’s not really about prompt engineering; it’s more about whether the tool can handle the specific requirements of consistency.

What skills are transferable across different AI image-generation tools?

Sachin Kamath: The skills that carry over are primarily visual skills—knowing what makes a good image in terms of composition, lighting, and overall aesthetics. This gives you an edge because you can communicate those needs through your prompts and better curate the outputs. The technical skill to use the tools is one thing, but having a strong visual eye, like professional photographers do, is what really elevates the quality of your results.

How important is prompt engineering in the future of image generation?

Sachin Kamath: While prompt engineering was essential, I think we’re moving toward a future where it’s less critical. AI tools are getting better at interpreting basic prompts, and more control is shifting toward using image references or visual inputs. Text prompts will still matter, but the visual direction will play a bigger role. Tools like Crea, where you control shapes and layout visually, are great examples of where things are headed.

How does AI speed up your creative workflow?

Sachin Kamath: AI tools can cut pre-production time by 80-90%. Where you’d normally need a week or more to prepare a creative pitch with storyboards and visuals, you can now do it in a day or two. With AI, you can visualize your ideas almost instantly, get client approval quickly, and move onto production with more confidence. It’s been a game-changer for creatives, especially in getting concepts approved faster.

What are some common use cases for AI-generated images among your students?

Sachin Kamath: Most of my students are creative professionals or leaders in design teams. They’re using AI for things like pre-production mockups, brand assets for social media, and even 3D modeling for product design. For example, a luxury furniture designer used AI to create high-quality renders of their concepts. AI tools speed up the ideation process and allow them to present polished visuals to clients much faster than traditional methods.

How do you stay up to date with the fast-evolving world of AI tools?

Sachin Kamath: It’s tough to keep up with the rapid evolution of AI tools. I block off two days a week to experiment and stay current. I follow news on social media, especially from companies I’m interested in, and I use my content creation time to explore new tools. I recommend blocking two to three hours a week for experimentation and bookmarking anything that catches your eye. It’s all about staying observant and giving yourself structured time to learn.

Check out the full interview right here ??

Product Spotlight: Consistent Character AI

Control and consistency are some of the biggest challenges when creating with AI. Sachin and his team have been truly obsessed with this problem and decided to release their own tool called Consistent Character AI. The tool goes beyond the visual look of a character and allows cretors to have also refined control on the pose unlocking many opportunities for creators.

You can try it out here: https://consistentcharacter.ai/

That’s it for today, and don’t forget to subscribe to the newsletter if you find this interesting

See you next week