Login to Flextreme AI

If you open the flextremeai.com site a Welcome page appears.
Scroll down and click on the Get Started button to open the login page.

You can use the app only with a Google account, but you don’t need to register.

After login, you see this simple and clean user interface and you can immediately start generating images:

Just type the prompt, select the aspect ratio and the number of images. Then click on Generate and Flextreme AI will do the rest for you.

Detailed, medium-length prompts also work. You don’t need to write the prompt in english. My experience is that Flextreme AI also understand german, italian and other languages. It’s worth to experiment with different languages.

Test #1 – Cinematic photo

Prompt: Cinematic photo of a red sportscar drifting in the sunset on Florida street. The sports car is taking a sharp turn. Low camera angle, full shot, motion blur, city lights, dust.

Test #2 – Landscape oil painting

Prompt: Oil painting of Swiss Alps in daylight.

Test #3 – Macro photo

Prompt: Extremely detailed photo of a beautiful blue tiger butterfly on a pink flower. Green background, shallow depth of field vivid colors, 4k fine art, sharp details.

Test #4 – Vector graphics

Prompt: Medium detailed and shaded vector illustration of a angry black cat warrior with shield and spear. He stands on his two back feets in the endless desert. Full shot, side view.

Test #5 – 3D Logo design

Prompt: A silver-cyan 3D gear logo on darkgray background with “COMPANY” written below the gear with silver sans serif font. A “YOUR SLOGAN HERE” text below the company text with serif font.

Test #6 – Using a complex prompt

My plan was to test the accuracy of the model and the prompt adherence. In fact, this is exactly the prompt from my Nano Banana Pro test:

Prompt: “Create a comic with the title: Getting started with Nano Banana Pro. Panel 1: A mobile interface on Gemini, rendered in a poetic watercolor-ink style with fine ink outlines and soft, bleeding washes. The interface is friendly and a hand painted with expressive brushwork taps a prominent button labeled “Create image”.

Above the button it should say “Choose your Model” then below there should be a checkbox that says “Thinking with 3 Pro” Muted greys and blues dominate the background. The button has a vivid yellow accent. “ Select the Thinking with 3 Pro model” and tap “Create image” to begin. Panel 2: A cheerful person is depicted adding their selfie from the phone’s camera. The user’s face is drawn with soft outlines and warm pastel colors, while the phone and UI maintain the delicate water-ink aesthetic. Visible paper grain adds to the texture.

Panel 3: The person thinks about what to create. In the background, different options are visualized to show what they’re thinking, including — them as a plushie, them with a mohawk hairstyle, and a figurine. These options are clearly rendered behind the person in the same style as the rest of the comic.

Panel 4: The person is shown adding a style prompt, with a speech bubble saying “Transform me into a watercolor painting”. The text is integrated into the panel’s watercolor-ink look, and the interaction feels natural and intuitive.

Panel 5: The person is seen editing the image by simply typing into Gemini. The scene captures the ease of this interaction, with the final edited image, now in a watercolor style, appearing on the screen. The overall tone is friendly, instructional, and inspiring. It feels like a mini tutorial comic, all conveyed through the specified delicate water-ink illustration style. Make the aspect ratio 16:9.”

Test #7 – 3D render

Prompt: 3D render of an enchanted treetop forest village with rope ladders on a summer evening with fireflies illuminating it during the blue hour

Test #8 – Pencil sketch

Prompt: A detailed pencil sketch of a flying dragon over the rocky mountains.

Troubleshooting

As I tested this app I experienced sometimes little but annoying bugs. If I have the solution, then I will share it here.

Issue #1 – Login service

Solution: just hit the refresh button or F5 key to reload the page. I’m sure that the Google login button should then appear immediately.

Issue # 2 – Strange results

Symptom: Strange text and irrelevant result.

Solution: Log out and log in again, refine your prompt.

Summary

Althrough the Flextreme AI art generator is still in alpha phase, it already works very good for basic tasks. In most cases the results are quite convincing. Below, I summarize my experiences in bullet points.

However, be aware that Flextreme AI does not save your generated images on their server.
Instead, they are storing them only in your browser session, as far as I understand. Therefore, I strongly recommend to download the images as soon as possible. Additionally, make sure you log in on the same computer or mobile device and use the same browser if you want to access them again.

Pros:

• easy to use interface
• fast image generating
• 15 images per day for free
• the minimum number of images is 1
• your generations are private

Cons:

• no pro plan / subscription
• images are for non-commercial use only
• minor application issues sometimes (eg. login)
• strange, irrelevant results sometimes
• no community gallery
• no extra features like:
  • negative prompt
  • AI art generation from reference image,
  • remove background,
  • inpainting,
  • selecting AI model
  • image enlargement

The post I tested Flextreme AI – Images and prompts are included appeared first on AI Art Blog.

How to try it out?

Nano Banana Pro is available in Google Gemini. You will see this starting screen when you open the Gemini webpage:

I would like to note that if you click on the Try It link at the top of the page, then Banana will load a sample prompt and generates an image, which already counts towards your 3 image limit per day.

Click on the Create Image button to activate the Nano Banana Pro image model.

Note: You can create only 3 images / day for free.

Test #1 – The inbuilt sample prompt

The prompt is very detailed and long, so I pasted it with a small sized font:

“Create a comic with the title: Getting started with Nano Banana Pro. Panel 1: A mobile interface on Gemini, rendered in a poetic watercolor-ink style with fine ink outlines and soft, bleeding washes. The interface is friendly and a hand painted with expressive brushwork taps a prominent button labeled “Create image”. Above the button it should say “Choose your Model” then below there should be a checkbox that says “Thinking with 3 Pro” Muted greys and blues dominate the background. The button has a vivid yellow accent. “ Select the Thinking with 3 Pro model” and tap “Create image” to begin. Panel 2: A cheerful person is depicted adding their selfie from the phone’s camera. The user’s face is drawn with soft outlines and warm pastel colors, while the phone and UI maintain the delicate water-ink aesthetic. Visible paper grain adds to the texture. Panel 3: The person thinks about what to create. In the background, different options are visualized to show what they’re thinking, including — them as a plushie, them with a mohawk hairstyle, and a figurine. These options are clearly rendered behind the person in the same style as the rest of the comic. Panel 4: The person is shown adding a style prompt, with a speech bubble saying “Transform me into a watercolor painting”. The text is integrated into the panel’s watercolor-ink look, and the interaction feels natural and intuitive. Panel 5: The person is seen editing the image by simply typing into Gemini. The scene captures the ease of this interaction, with the final edited image, now in a watercolor style, appearing on the screen. The overall tone is friendly, instructional, and inspiring. It feels like a mini tutorial comic, all conveyed through the specified delicate water-ink illustration style. Make the aspect ratio 16:9.”

And this is the result:

Click on the image to enlarge

It’s pretty convincing. The typography is perfect. I don’t see any typos or other problems on the picture.

Remark: Gemini also included the banana emoji in the prompt, but I had to delete it to able to save the blogpost.

Update: I tried the same prompt with ChatGPT (GPT Image 1.5):

This results is impressive too. I think that Nano Banana Pro has a new rival.

Test #2 – Infographic

I used this prompt:

Create an infographic-like image of the most popular brushes in similar style as on the attached example picture, but with landscape aspect ratio (eg. 16:9). The brushes are from top to down: 2B Pencil, Ballpoint Pen, Soft Airbrush, Watercolor brush, Acryl brush, Oil paint, Calligraphic brush. Each row contains the stroke of the brush or tool with realistic and relevant color. Next to the stroke is the photorealistic illustration of the tool (for example it is clearly visible that the brush stroke was painted with the brush) and a sample image that was drawn with the tool. The image should also contain the name of every tool with a good readable darkbrown coloured font. There is also a darkbrown coloured victorian ornamental frame around the infographic.

The reference image was Krita’s first picture about brushes.

This is the generated image:

The result is impressive! Gemini precisely followed my instructions: text, tools, samples, structure, victorian border: everything is on the picture.

Using Banana in Leonardo.Ai

You can also try it in Leonardo.Ai, but in the free plan you have only 150 tokens. One image with Nano Banana Pro costs 140 tokens, so you can generate only 1 image.
Tokens are resetting in 12 hours, so you have a 2 image daily quota.

In Leonardo free plan you can use 1 reference image. I made a test with it.
This is the starting real estate interior photo:

This is a very simple prompt to replace some objects on this living room photo:

Replace the pictures on the wall with photos from Gobi desert. Replace the couch with a darkbrown leather couch. Change the wall colour to matching bluegreen, increase the density of the foliage.

And this is the final result:

I think it’s pretty good. Nano Banana also was able in August to make this kind of change on a picture – I think.

The best thing is, that it doesn’t change the alignment or form of the objects in the scene, it changes only what you define in the prompt.

Let’s summarize what are the new features of the 2nd version of Nano Banana.

Summary

Nano Banana Pro is a revolutionary model. It’s fantastic and a little bit scary at the same time. Images made with this new model are more realistic and hard to make a difference between real photo an AI generated image. In another article I will post images which compare the standard and pro version of Banana.

The main new features are:

• enhanced prompt adherence
• better consistency (you define what to change or replace, the other objects and forms remaining intact)
• ability to create complex text and image combinations on one picture, like:
  • infographics
  • comics
  • accurate scientific graphics
  • easily change the mood, like: lighting, season, camera angle etc.

The post I tested Nano Banana Pro – Check the Results appeared first on AI Art Blog.

Pixlr AI image generator gives you 20 free credits, which is enough for 20 image generations. Every time you hit the Generate button, 4 images will be generated. It costs 4 credits (1 credit per image).

I used some basic prompts. These prompts were based on those written in the Pixlr newsletter. Of course, I changed them a bit to make the end result varied.

Image properties:

• Aspect ratio: 1:1
• Image dimensions: 1024 x 1024 pixels
• File format: .png

I used the default settings (No Style, No Color, No Lighting, No Composition) and I gave only the prompts below and after that I clicked the Generate button.

You can view the full size test images on my Facebook page.

Test #1 – Dragon on the cliff

Prompt: A dragon sits on a rock and overlooks a huge medieval city. Use deep greens and yellows for the scales and make sure the city below reflects the past.

Result: Beautiful, detailed, 2D painting style dragon, rich in colors. Maybe he needs some color adjustment. Drawing a dragon is not an easy task because of its complexity.

The last image needs some retouch (separate castle at the bottom right corner)

Test #2 – White Wolf

Prompt: A white wolf with shiny iridescent fur standing in a mystical moonlit forest with millions of tiny stars shining in the sky.

Result: interesting, dreamlike, mystical wolf images. A little too much purple and pink in the generated images imho. I would like it better with a bluer color scheme. My favorite is the second creation.

All wolves are four-legged and I did not use negative prompts.

Test #3 – Zombie silhoutte

Prompt: A silhouette of a zombie approaches the viewer on a full moon night in an eerie, fog-shrouded forest. Pale bluish color scheme.

Result: I like the color tone and the overall mood of the images. It was a good idea to describe the colors in the prompt. The second picture turned out to be scary enough. The last image has a strong, centered composition. Zombie eyes can be made more dramatic with a small brush in an image editing software.

Summary

Pixlr AI Image generator is a pretty interesting app with good image generation capabilities. The images can be customized prior to the generation with styles, color tones, lighting and composition. These customization techniques was not the topic of this article. I will test the custom features maybe later.

Tip: You have also the opportunity to post process the image. Just click on the paint bucket icon and then choose the AI Tools from the left toolbar.

Note: All AI tools costs additional credits.

As I scrolled the gallery on the Pixlr site I noticed that many of the generated images have anomalies like humans with 6 fingers, not physically accurate electronic devices, unwanted objects or inaccuracy in texts. These are other areas where Pixlr can make further enhancements.

Notice: I don’t tested the negative prompt feature which maybe can solve some of this common problems.

What you think about this tool? Share your thoughts and experience in comments!

Thank you for reading this article! Don’t forget to check other AI image topics!

The post AI Image Generator Test – Pixlr appeared first on AI Art Blog.