I mostly mess with Midjourney, but recently revisiting Tenser just for fun for Halloween creating monsters because it's more NSFW friendly for Horror stuff. I now feel Tenser is capable than a lot more than I realized after making images with several LORAs combined. I'm fascinated with the idea of making my own now and just curious if the background of the pictures affect it. Does it need to be a consistent plan background or is this good and sticking to the subject? Should you take pictures with different expressions for reference or does that not matter? Enjoying this content, again feel like I underestimated this tool or maybe the LORA library has blown up since I used it before.
I am a noob at this tuff so I have a few questions: 1) If I download the Lokr or Lora I trained, can I use it on Forge UI? 2) Can the images we generate be used commercially?
@@JonathanHubJam Oh, yes! And I'm actually using Nvidia Broadcaster to do the actual swap. It only works with Nvidia RTX cards - and I'm really only using it because currently my studio configuration doesn't let me put up a proper green screen, so this is an AI removal, and it does a pretty good job I think...but I'd prefer my green screen.
...It seems I can redeem 100 credits from your page, but extra instructions may be needed. May be other methods, but for me it worked with pressing the "Follow" option and then register. Thank you and hopefully you also got some referral credits for it!
I have 49 hand selected super cute images of original 3d anthro furry creatures all in the same style, but almost all of different kinds of creatures from insectoid to lizards and mammals, most generators absolutely suck at creating this kind of thing so this will be brilliant. I'm uploading the images now! I changed epoch setting to 2, I'm not sure what any of the settings do but I hope it will help. Hope I don't need to change all the other settings, if anyone knows how to get the most out of this please help because I have 3100 credits and I don't want to waste them or time on multiple trainings. Thanks a bunch
Good to know it works well for those type of images lol. After training the Lora on the next step called "Host a Model" how many training steps did you put, and what does it mean? Bob used 220 but he didn't explain that part unless I missed it.
I did not test yet (just saying you can *wink). You can get better results with more steps and so, I guess, depends on your expectations. But I think most result will be good enough. - I would train a Flux model or get a SD1.5 (or XL) to test it my own, Automatic1111 or Comfuyi.
Great surprise at the end! Loved it
I mostly mess with Midjourney, but recently revisiting Tenser just for fun for Halloween creating monsters because it's more NSFW friendly for Horror stuff. I now feel Tenser is capable than a lot more than I realized after making images with several LORAs combined.
I'm fascinated with the idea of making my own now and just curious if the background of the pictures affect it. Does it need to be a consistent plan background or is this good and sticking to the subject? Should you take pictures with different expressions for reference or does that not matter?
Enjoying this content, again feel like I underestimated this tool or maybe the LORA library has blown up since I used it before.
I am a noob at this tuff so I have a few questions: 1) If I download the Lokr or Lora I trained, can I use it on Forge UI? 2) Can the images we generate be used commercially?
did you record the video in another language then use ai for voice and lip sync?
Can this be done on Mimicpc?
Thanks for the extra credits! Looks good.
I see there’s a TensorArt app for iOS. Can i train that also and all these steps how you do here in the App Sir?
Great one Bob, just trained myself on LOKR :) Lookin' smoking hot
Hi, do you achieve the background swap with OBS? Or is it a different software? Thanks.
Sorry, which background swap are you talking about exactly?
@@BobDoyleMedia I mean in all your videos when you're talking to the camera - that studio background is swapped in isn't it?
@@JonathanHubJam Oh, yes! And I'm actually using Nvidia Broadcaster to do the actual swap. It only works with Nvidia RTX cards - and I'm really only using it because currently my studio configuration doesn't let me put up a proper green screen, so this is an AI removal, and it does a pretty good job I think...but I'd prefer my green screen.
I think you put the wrong link in the description? It leads to your profile page on Tensor Art.
...It seems I can redeem 100 credits from your page, but extra instructions may be needed. May be other methods, but for me it worked with pressing the "Follow" option and then register. Thank you and hopefully you also got some referral credits for it!
@@CitadelAi yes i will help you, AFTER clicking on bob link, sing in into TENSOR ART, then go to HOMEPAGE then TRAINING..... then do all the stuff.
great content! Thank you for your good work
Is there’s a option to choose other formats as 9:16 or other?
@@rashid4players that all happens when you generate the image. It’s not really part of this process.
@@BobDoyleMedia I don't quite understand that because the options are Square Landscape and Portrait. There stands not other options.
Thanks for sharing, great explanation!
I was literally just b watching you and you pulled up
Appreciate the knowledge 🙏
bob, can i load this LOKR as a LORA on FORGE UI flux dev? or how do i load it?
@@liquidmind yes, I should’ve made that clear. It loads just like a Lora.
@@BobDoyleMedia sorry mate, you give so much info, i missed it!!! lol
@@BobDoyleMedia i am having issues with the PAYMENT to make more training..... bummer....
So can we download this like a model? Like all this work to spend more or can we use this in sdxl or anything else???
Nevermind you said it
how many credits did u use for 10 images training?
11 = 198
You said New York City u must be New new Yorker I will look for you😂
You'll be looking a long time... 😁
I have 49 hand selected super cute images of original 3d anthro furry creatures all in the same style, but almost all of different kinds of creatures from insectoid to lizards and mammals, most generators absolutely suck at creating this kind of thing so this will be brilliant. I'm uploading the images now!
I changed epoch setting to 2, I'm not sure what any of the settings do but I hope it will help. Hope I don't need to change all the other settings, if anyone knows how to get the most out of this please help because I have 3100 credits and I don't want to waste them or time on multiple trainings. Thanks a bunch
Oops it's going to take 16hrs with 24 steps and 10 epochs
How did you get that many credits on TensorArt?
I am a little suspicious of your professor, though. May be your smart cousin? :)
I'm quite sure I have no idea what you're talking about...😏
26 credits to train a model? it costs 252. Otherwise, thanks for excelent tutorial
The credits depend on how many images you chose, how many steps, etc. Thanks so much for watching!
I subscribed! Don't look for me!
it is really easy, guys, I made her wear nothing. lol
Good to know it works well for those type of images lol. After training the Lora on the next step called "Host a Model" how many training steps did you put, and what does it mean? Bob used 220 but he didn't explain that part unless I missed it.
I did not test yet (just saying you can *wink). You can get better results with more steps and so, I guess, depends on your expectations. But I think most result will be good enough. - I would train a Flux model or get a SD1.5 (or XL) to test it my own, Automatic1111 or Comfuyi.
no it doesnt work it just spit error
Train failed!
rpc error: code = Unknown desc = network_rank_dim must be less than or equal to 128