I think what most of us want to know is will this work to pose a character in many sorts of different poses. Action poses. Can you make Lara leaping towards the viewer with sword in hand. Or have her smashing through a sky light Batman style. This is the kind of stuff people want to do n
@scenario_gg Hi, If i trained lora on images that do not have certain emotions and facial expresions, will it still be possible to get those emotions with your charachter while generating with your lora?
Thanks for making these, but can I humbly suggest slowing them down significantly please. Once you know how to do something, it all seems "obvious" but until that point, every single step is a potential sticking point. Thanks :)
There may be a better solution, you could just use what is available: The video speed can be slowed down significantly (1), and the subtile gives you a visual documentation of what is said (2) (of course you probably would mute slower videos), and lastly just hit space key/tap screen ONCE and take the time you need what´s the problem? It will however be problematic for the content creator.. No one would bother to take the time and watch lengthy tutorials , which makes the last point even worse. Do you know how many hours of work go into even 1-2 min video clips? A significantly increase would mean you would almost work a full day just to get one video out, that just gets ignored.
@us_f4rmer Yes, they would take the time to watch it. That's why they're there. Especially if it's regarding something they're trying to make a living from. And it can mean more content by breaking the vids up into a series. And as another person commented, slowing it down degrades the audio. The more useful a thing as is, the more it is used (or in this case watched).
I want to know how can I upload an image from outside this platform and make a skin system character? How can I bring in the characters that I drew into a I
We used the picture to ask chatgpt for a prompt that describes the character in it. Then we copied the prompt we got from chatgpt and added it to the original prompt we had written to generate the character. This ensures we get a consistent image dataset to use for the training.
I did the training with some of my pictures. So now if I select the model I have created, and I clic on "Generate", it generates 4 pictures of a man that looks like me. Not bad. But this happens only if don't write any prompt. If I write a prompt, such as "A man having lunch", it creates 4 pictures of a man having lunch, but it doesn't look like me at all.
You have the option to train multiple models and duplicate your model to experiment with different training methods for comparison. You can also modify the training data by adding or removing images and then retraining the model.
@@scenario_gg Would you mind telling me real quick if using unity would allow me to fetch model files? I think I read something to this effect somewhere, but there's a good chance I got that wrong. I mean, does the choice of words necessarily imply that there WILL be a specific moment when we can? 🤔🤨
I want to take my photograph and make the templates you provide, cannot figure it out, and its annoying you want to sell this thing and I don't want to spend hours learning something that does not do what I think it can
Hello, i saw a man reup your video on tiktok and make a how to video by his self. It completely wrong turtorial. I reported him on Tiktok but nothing happen. You need to do something.
The ChatGPT step is bizarre and useless, why not use CLIP or Deepboru for SD, chatGPT is doing something similar, but it's steps removed from SD... just... why?
I think what most of us want to know is will this work to pose a character in many sorts of different poses. Action poses. Can you make Lara leaping towards the viewer with sword in hand. Or have her smashing through a sky light Batman style.
This is the kind of stuff people want to do n
@scenario_gg Hi, If i trained lora on images that do not have certain emotions and facial expresions, will it still be possible to get those emotions with your charachter while generating with your lora?
Thanks for making these, but can I humbly suggest slowing them down significantly please. Once you know how to do something, it all seems "obvious" but until that point, every single step is a potential sticking point. Thanks :)
Noted!
There may be a better solution, you could just use what is available: The video speed can be slowed down significantly (1), and the subtile gives you a visual documentation of what is said (2) (of course you probably would mute slower videos), and lastly just hit space key/tap screen ONCE and take the time you need what´s the problem? It will however be problematic for the content creator.. No one would bother to take the time and watch lengthy tutorials , which makes the last point even worse. Do you know how many hours of work go into even 1-2 min video clips? A significantly increase would mean you would almost work a full day just to get one video out, that just gets ignored.
@@us_f4rmer Slowing down degrades audio quality more than speeding it up :)
@us_f4rmer Yes, they would take the time to watch it. That's why they're there. Especially if it's regarding something they're trying to make a living from. And it can mean more content by breaking the vids up into a series. And as another person commented, slowing it down degrades the audio. The more useful a thing as is, the more it is used (or in this case watched).
I want to know how can I upload an image from outside this platform and make a skin system character? How can I bring in the characters that I drew into a I
did you use the chatgpt 30 word description and the picture as a reference to generate the images that were used to train the model?
We used the picture to ask chatgpt for a prompt that describes the character in it. Then we copied the prompt we got from chatgpt and added it to the original prompt we had written to generate the character. This ensures we get a consistent image dataset to use for the training.
but how do you paste/upload the image into chatgpt? there is no way to do that, as far as i can tell@@scenario_gg
I did the training with some of my pictures. So now if I select the model I have created, and I clic on "Generate", it generates 4 pictures of a man that looks like me. Not bad. But this happens only if don't write any prompt. If I write a prompt, such as "A man having lunch", it creates 4 pictures of a man having lunch, but it doesn't look like me at all.
I’ve come across this too. Some extra step is missing.
Do we get same character, or “looks alike”, can I train model and generate images for graphic novel?
You have the option to train multiple models and duplicate your model to experiment with different training methods for comparison. You can also modify the training data by adding or removing images and then retraining the model.
Can you download you generated model from Scenario to be used in other SD applications? Or is the model only usable on your platform?
Hi, for the moment models are only usable on platform :)
@@scenario_gg Would you mind telling me real quick if using unity would allow me to fetch model files? I think I read something to this effect somewhere, but there's a good chance I got that wrong.
I mean, does the choice of words necessarily imply that there WILL be a specific moment when we can? 🤔🤨
Please make a tutorial about posing and controlnet inside Scenario, I'm having a lot of difficulty making it work
Hi, in my promt I put "full body" and it keeps drawing half body shots, even adding "half body shot" to the Bad promt. Any solution?
Try “I want to see this character in a 360 degree view”
Thank you
Just FYI that I could only sign up with my phone; your website just kicks me back to putting in an email to get started.
any suggestions for that 30 word description outside chatgpt?
It might not be helpful as it also requires a subscription, but Midjourney has a Describe feature for uploaded images.
I want to take my photograph and make the templates you provide, cannot figure it out, and its annoying you want to sell this thing and I don't want to spend hours learning something that does not do what I think it can
Hello, i saw a man reup your video on tiktok and make a how to video by his self. It completely wrong turtorial. I reported him on Tiktok but nothing happen.
You need to do something.
The ChatGPT step is bizarre and useless, why not use CLIP or Deepboru for SD, chatGPT is doing something similar, but it's steps removed from SD... just... why?
It's way too confusing for a beginner.
hi, this video wasn't useful at all. I thought it was needed to create a Lora model, but no