This guide is so insightful! Fine-tuning AI models can feel like trying to solve a puzzle-every piece matters! I'm super curious about how tools like SmythOS can simplify this process.
Thank you so much! And... good point! So far, I have been using crewAI for agents. But I'm trying to explorer other options. I'll definitely give SmythOS a try ☺️
TY. You are producing some great tutorials. TBH the final girl character didn't really match the reference images. No freckles, looked younger, mouth, nose and other features were very different. Is there a way to achieve a more accurate resemblance? Maybe use more reference images?
Thanks so much for your great comment! And yes, you are right. Using image-to-image or "image guidance" as Leonardo now calls it, you can achieve more resemblance. In case you missed it, here my video on the new image guidance feature: th-cam.com/video/-7-yvan5IbA/w-d-xo.html.
Can this be done with a food product? For example, you train AI with different pictures of chocolate bar with your name like Zozic and ask Leonardo to generate images with it? For example, superman eats Zozic.
Thanks for reaching out! While I haven't tested AI with food images, your idea seems like a fun experiment. It should work in theory. If you give it a shot, I'd love to hear how it goes. Drop your results in the comments!
Thanks for the reply, yes, it is interesting and I tried it with 20 images of potato chips bags but the result wasn’t good at all. It seems that AI didn’t understand my prompt and generate something wrong. Sometimes it generated chips bag but was different with data set and not in standard size. I hope you try it and probably you will achieve better result. @@TechXplainator
mine isnt working like yours. it takes me back to the image generation page and it all looks the same. I also don't have a negative prompt I can put in.
After which step does it take you back to image generation page? When you go to "Finetuned models" then "Your models" you should see your fine-tuned models. Don't you see anything there? Regarding negative prompts, there's a slider that you must activate to add a negative prompt. You'll find it right next to the "Elements" under the prompt box. I hope this helps ☺️
I'm sorry to hear that. I've been considering creating a video about what to watch out for when fine-tuning a model in Leonardo. But I haven't done it yet because I wasn't sure if people would find it helpful. Would you be interested in something like that?
sure, my prompts were actually very simple: ozan-girl with purple hair ozan-girl as a red witch ozan-girl with glasses ozan-girl as batman I'll also add them into the description of the video for others to see. Thanks for pointing that out :-)
Thanks this was good until I downloaded the images, resulting to Canva hiding them behind some mysterious greedy mode and this Canva does not have help. So much for using Canva.
Hey! Sorry to hear you're having trouble with Canva, it worked well for me although I'm using the free plan. Even though I don't work for Canva, I'd suggest reaching out to their support for help: www.canva.com/help/contact-us/. They might be able to sort out the issue. Hope you get it sorted soon! 😊
This guide is so insightful! Fine-tuning AI models can feel like trying to solve a puzzle-every piece matters! I'm super curious about how tools like SmythOS can simplify this process.
Thank you so much! And... good point! So far, I have been using crewAI for agents. But I'm trying to explorer other options. I'll definitely give SmythOS a try ☺️
Very helpful and concise video. Thanks!
Another excellent video... your channel is very good! i'm sure it's going to be a giant channel!!!
Thank you so much! I hope so too!
Thank you so much for the helpful and informative tutorial!
❤❤❤❤❤❤
TY. You are producing some great tutorials. TBH the final girl character didn't really match the reference images. No freckles, looked younger, mouth, nose and other features were very different. Is there a way to achieve a more accurate resemblance? Maybe use more reference images?
Thanks so much for your great comment! And yes, you are right. Using image-to-image or "image guidance" as Leonardo now calls it, you can achieve more resemblance. In case you missed it, here my video on the new image guidance feature: th-cam.com/video/-7-yvan5IbA/w-d-xo.html.
Thanks u bery much. ❤
Can this be done with a food product? For example, you train AI with different pictures of chocolate bar with your name like Zozic and ask Leonardo to generate images with it? For example, superman eats Zozic.
Thanks for reaching out! While I haven't tested AI with food images, your idea seems like a fun experiment. It should work in theory. If you give it a shot, I'd love to hear how it goes. Drop your results in the comments!
Thanks for the reply, yes, it is interesting and I tried it with 20 images of potato chips bags but the result wasn’t good at all. It seems that AI didn’t understand my prompt and generate something wrong. Sometimes it generated chips bag but was different with data set and not in standard size.
I hope you try it and probably you will achieve better result.
@@TechXplainator
I love you ❤! Thank you so much!! You are a angel send from God himself!
Wow, thank you! I'm so glad you enjoy my videos! Let me know what topics you want me to show more of ☺️
great
mine isnt working like yours. it takes me back to the image generation page and it all looks the same. I also don't have a negative prompt I can put in.
After which step does it take you back to image generation page? When you go to "Finetuned models" then "Your models" you should see your fine-tuned models. Don't you see anything there? Regarding negative prompts, there's a slider that you must activate to add a negative prompt. You'll find it right next to the "Elements" under the prompt box. I hope this helps ☺️
It works now with no upload images. But when I go to generate pictures, they turned out all disfigured and blurry. Pretty bumbed@@TechXplainator
I'm sorry to hear that. I've been considering creating a video about what to watch out for when fine-tuning a model in Leonardo. But I haven't done it yet because I wasn't sure if people would find it helpful. Would you be interested in something like that?
I was missing the character part. Can you share the prompts you tried for the girl?
sure, my prompts were actually very simple:
ozan-girl with purple hair
ozan-girl as a red witch
ozan-girl with glasses
ozan-girl as batman
I'll also add them into the description of the video for others to see.
Thanks for pointing that out :-)
Thanks this was good until I downloaded the images, resulting to Canva hiding them behind some mysterious greedy mode and this Canva does not have help. So much for using Canva.
Hey! Sorry to hear you're having trouble with Canva, it worked well for me although I'm using the free plan. Even though I don't work for Canva, I'd suggest reaching out to their support for help: www.canva.com/help/contact-us/. They might be able to sort out the issue. Hope you get it sorted soon! 😊
Great tutorial!✅