This is great but I have some questions: 1) when we browse civitai should we stick to models that work for 1.5 or is Krita adaptable to new versions? 2) are we required to add the lora prompt Keywords all the time in our prompts? What's then the purpose of downloading a lora that makes, I don't know, bearded vikings, if we must add "bearded viking" as a keyword all the time? And negative prompts too? 3) I sadly got a little lost with the VAE, I understood the checkpoint you downloaded was missing something important but that s as far as I got. I'm still very grateful for your videos and will keep following you, if possible I'd love the next to be on keeping character consistency in subsequent generations, this is what got me into checkpoints and Loras to start with. Thanks again for the great content.
1) SD1.5 model on civitAI has an improved/Refined version which I showed at 5:32 horizontally they will remain with base SD1.5 Is Krita Adaptable to new versions - That will be SDXL (new version) model, it needs a different workflow I installed at 3:47 So, yes Krita adapted. 2) We do not require to have the lora in prompt with the current version. Using a vikings lora with a strength of 100% will give you the character 'viking' even if you do not use the trigger words. Let's assume you are using a clothing LoRA, the LoRA here is capable of formals and traditional wear. Now you have to specify the trigger words based on how you want the character? In formal office suit or the old traditional style. If you do not specify, lora will be free to do based on the prompt and weight. Negative Prompt - SD1.5 and SDXL gives bad images sometimes. Specifying words like deform / nude will avoid those things happening. You must use negative prompt always when working for SD. 3) The VAE was included in the first model '3DCartoon' I used, it was also mentioned by the owner of the model showed at 16:33 The second checkpoint did not have a VAE included in the checkpoint, Hence I downloaded from a different pace and kept it inside VAE folder.
![แค่บอก - Atom ชนกันต์ [OFFICIAL MV]](http://i.ytimg.com/vi/kgwhtg8bAxY/mqdefault.jpg)
This is great but I have some questions: 1) when we browse civitai should we stick to models that work for 1.5 or is Krita adaptable to new versions? 2) are we required to add the lora prompt Keywords all the time in our prompts? What's then the purpose of downloading a lora that makes, I don't know, bearded vikings, if we must add "bearded viking" as a keyword all the time? And negative prompts too? 3) I sadly got a little lost with the VAE, I understood the checkpoint you downloaded was missing something important but that s as far as I got. I'm still very grateful for your videos and will keep following you, if possible I'd love the next to be on keeping character consistency in subsequent generations, this is what got me into checkpoints and Loras to start with. Thanks again for the great content.
1) SD1.5 model on civitAI has an improved/Refined version which I showed at 5:32 horizontally they will remain with base SD1.5
Is Krita Adaptable to new versions - That will be SDXL (new version) model, it needs a different workflow I installed at 3:47
So, yes Krita adapted.
2) We do not require to have the lora in prompt with the current version.
Using a vikings lora with a strength of 100% will give you the character 'viking' even if you do not use the trigger words.
Let's assume you are using a clothing LoRA, the LoRA here is capable of formals and traditional wear.
Now you have to specify the trigger words based on how you want the character? In formal office suit or the old traditional style.
If you do not specify, lora will be free to do based on the prompt and weight.
Negative Prompt - SD1.5 and SDXL gives bad images sometimes. Specifying words like deform / nude will avoid those things happening. You must use negative prompt always when working for SD.
3) The VAE was included in the first model '3DCartoon' I used, it was also mentioned by the owner of the model showed at 16:33
The second checkpoint did not have a VAE included in the checkpoint, Hence I downloaded from a different pace and kept it inside VAE folder.
ltx install google colab. Please creating video I am waiting 😢
hmmmmm