Subscribe
Sign in
AI Image Transformer — Transform your pictures into cartoons or custom styled arts
AI Image Transformer - quickly turn your pictures into cartoons or custom styled arts using pre-trained AI models running on the Oasis blockchain. Simply upload your images to process, and download the results shortly after, absolutely hassle free.
37
Replies
@tjerkw There seems a bug...working on it now. The Neuralcanvas looks like a specialized stylizer, let me know how you want to cooperate.
@tjerkw The bug is fixed, but the stylizer does take extra time to return the results, so I also increased the timeout values. Be patient and you'd get the results, and try to upload to process again if you get some unexpected response.
Hey buddy, I think you are leaking photos from another users on your interface. After trying a few times I started to receive back images that I didn't uploaded myself...
@igor_escobar Yes it's a bug. I just updated the server and restarted it, please refresh and see if the problem still happens.
The advance in computer vision makes it possible for anyone to turn an image into a cartoon, or transfer it to an artistic one by applying the style of an art work. But normally AI models are complicated to set up and run thus unavailable to many people. The AI Image Transformer makes use of the scalable and secure computing environments of Oasis Network (a privacy-enabled blockchain platform for open finance and a responsible data economy), it aims to give users around the world a simple interface to cartoonize and stylize images using pre-trained AI models, and the result is to be produced in near real-time. It's really simple to use, just upload your images and download the results. Have a try and see if you like it.
Hello fellow makers, thanks for your interest in the app and kind words about it. @defertu, I don't have a clear idea about the commercial use yet, but it'd help if you can send a proposal to info@aifinancials.net.
Have you ever wished to use cartoonized pictures of yourself in your online profiles? or wondered what the real world pictures look like as cartoons? The Cartoonizer can give you the answer quickly.
Have you ever dreamed of great styled arts created by yourself after visiting an exhibition? and do you want to see what the photos you took would look like should they be worked on by the great artists? The Stylizer can give you the answer quickly.
Check out the video for the app at https://youtu.be/cWG22XKiW3A, enjoy!
The results generated by the app can be easily minted as NFTs. As an example I created a collection on OpenSea: https://opensea.io/collection/tr.... There's no upfront cost or gas fees, but OpenSea will take 2.5% for every NTF sold. So if you get some creative images and want to list them as NFTs, you can also do that now manually, with a crypto wallet. I plan to create a NFT bridge to simplify the task, or even automate the whole process. At which point everyone could turn pictures to arts and list them as NFTs in one or two clicks. Please share what you think.
Sketch is a simple form of art and anyone may have tried it at some point. Turning photos into sketches is fun and can save a lot of time than producing the drawings by hand. The Sketcher is the latest function added to the AI Image Transformer app, it takes one input image and produces three sketches of different styles and two interesting byproducts for detecting the edges and tone of the picture.
WHAT'S NEW
- Enhanced the Stylizer with "Preserve Background" option, when enabled, the Stylizer will try to detect the background area in the source image and apply the style to the foreground only. This can make the resulting images more appealing when the source image has a relatively large background area.
- Reworked on the list of predefined style images of the Stylizer to include more interesting styles, that may generate unique digital arts for NFTs.
WHAT'S NEW
- Added new functionality: Image Multi-Stylizer, that can take a source image and one or more style images, and use a semantic-aware neural network that automatically incorporates multiple styles based on the regional semantics.
- Added new functionality: Image Enlarger, that can upsize images at a scale of up to 8 (64 times of the original size).
Note: In preparation for the launch of a NFT portal on the Oasis Network (a versatile blockchain with low fees), the generated images of Stylizer, Multi-Stylizer and Enlarger will contain watermarks by default. Soon you will be able to pay with cryptocurrency to have the desired images generated without watermarks, and use them to mint NFTs or for other purposes.
WHAT'S NEW
Added support to batch parallel processing for all functions, now in Cartoonizer, Sketcher, Enhancer and Enlarger you can select multiple images to process, and in Stylizer and Multi-Stylizer you can select one source image and generate results for every predefined style or some combination of styles in one click, which will make it much simpler to identify the styles that are particularly interesting for a certain source image.
WHAT'S NEW
- Improved the progress bar after clicking "Upload to Process" button, to make it smoother and reflect closer to the overall progress of each function.
- Enhanced error handling, to make the functions especially batch ones more reliable despite the unpredictable blockchain status.
I'd like to share a major update about the AI Image Transformer app, which has added a new function: Image Enhancer, that can improve image quality with increased sharpness and more vivid color but preserve the tonal styles.
Tried with one image. Liked it. Give me an option to change the intensity of effect.
Congrats on the launch.
Took a lot time to upload a normal quality photo but was surprised with the result. Loved my cartoonized version. Keep up the good work.
@rcrrr_poudel Glad you like it. It's probably the processing not uploading that took much time, as explained on the welcome page: Note the computation units on the blockchain are started on demand (not always on), so although the AI model itself doesn't take long (a few seconds) to process an image, a request that happens to start a new instance can take a lot longer (tens of seconds) to get the result.
Neuralcanvas