From the very beginning of the ChatGPT era, I made it a habit to stay updated with the latest AI developments. Things became even more exciting with the emergence of MidJourney. I still vividly remember the day I first heard about it — it felt almost impossible to believe! I wondered how it could be possible to generate images using just a few keywords.
As days, months, and years went by, AI evolved, becoming more powerful and refined. I began incorporating these advancements into my professional work. For every conceptual graphic project, I now rely heavily on “Ideogram.”
The introduction of Copilot with the Windows 11 update made things even more convenient, with AI tools just a sidebar away. With each new update, these AI models continue to become smarter and deliver increasingly accurate outputs.
However, I remained curious about Google’s progress in this space. When they released Gemini, I felt it fell short compared to other models like Copilot and Claude. I eagerly awaited the launch of Google’s AI image generator, and after much anticipation, the day finally arrived — Google released Imagen 3 last month.
With this milestone, I decided to put various AI platforms to the test, comparing how each one performs when generating images based on the same prompt.
The prompt
The image features a cartoon character that resembles a small cat or cat cub. The character has large,expressive eyes that give it a cute and innocent appearance. The cat is wearing a black shirt and black pants. It has a black mask-like design around its eyes, possibly indicating a superhero theme. The character is smiling and walking confidently with one foot forward. Color and Lighting: The cat’s fur is brown, with a darker mane around its head. The eyes are brown with a glossy, reflective quality. The scene is set with soft, warm lighting coming from the left side, casting shadows on the floor. The background is dimly lit, suggesting an evening or night-time setting. Setting: The character is indoors, walking towards an old-fashioned TV on a small wooden table. The TV has a vintage design, with a curved screen and dials on the side. The room appears to be cozy, with minimal light, emphasizing the character as the focal point. Environment: The floor is wooden, with a smooth, reflective surface that captures the light from the TV. The background includes a window or some kind of opening on the right side, through which some light is faintly coming in. There is a subtle texture on the walls, with a dark and muted color palette. Emotional Tone: The overall atmosphere is warm and friendly, likely evoking a sense of nostalgia or comfort. The character’s expression and posture suggest determination or excitement, possibly implying that the character is on a mission or about to engage in an adventure.
The results are below.
Google Imagen ImageFx provides four generations per entry. Comparing as a new model, I found the outputs are satisfying and pretty much detailed.
Dall-e provides one generations per entry. Comparing as a old and updated model, I found the outputs are not top notch and there is room for more development.
Ideogram provides four generations per entry. Ideogram is my first choise for image generation for its super fast and perfect output. The cool thing is we can use the new version for limited time as well.
Hail Ideogram ❤️
Microsoft Co-pilot provides four generations per entry. If I need to give comments on this engine I should say, Good for daily usage but never expect perfect result.
I do not know when Perchance released but I found it recently. Perchance provides six generations per entry. You can have different kind of filter for the results. Outputs are “ok”.
The most interesting thing is it is uncensored! You can create anything you want. You know what I mean, right? 😜
This one is probably new. Just one option per generation. I personally do not like the outputs at all. You can see the result above.
I know there are more AI image generators right now. But I tried to focus on the popular and free models available there.
I would love to have your feedbacks.
And don't forget to comment or reach me if you have any queries regarding art and architecture and the latest ai trends.
Use of Generative Artificial Intelligence (AI) and AI-Assisted Technologies
During the preparation of this article ChatGPT and MS Copilot was used to improve readability and language.
Keywords: generative ai, google imagen 3, ideogram, dall-e, Lime wire, Co pilot
Commentaires