Charmingcontours

TRAINING GUIDEIn previous article we learn how to train your own model on Flux.This time I'll share with you my experience creating a Illustrious model on-site training feature. After many models trained I finally found the best configuration settings that worked for me and I hope it works for you as well.First thing firstTensor.art AccountSet up an account on Tensor.art and ensure it has a minimum of 700 credits.Dataset GuidelinesFor optimal results, it is advisable to include between 10 - 30 images in your dataset. The majority of these should be close-up photographs, complemented by a few mid-range shots and at least one full-body image. Additionally, incorporating side profile images alongside front-facing ones will enhance the dataset's diversity.Here is my dataset that I've used for this LoRa.Guidelines for Captioning PhotosTensor.art automatically put captions on your images but consider adding:Unique Trigger Word: Begin each caption with a distinctive trigger word that identifies the dataset.Also consider this on your captions.Descriptive Tags: Include tags for all elements you wish to vary in the generated images. For instance, if you intend to modify attributes like hair color and length, ensure these are detailed in every photo caption.Consistent Elements: Any aspect not explicitly described and that remains consistent across all photos in the dataset will automatically be associated with the trigger wordParameter Settings and Likeness CaptureUtilizing a collection of approximately 10 portrait photographs, a highly effective and adaptable likeness model can be developed.Parameter Guidelines• Higher rates = stronger character features but potential loss in image quality• Lower rates = better image quality but weaker character features• Most character models work well with UNET around 0.0005 and TE around 0.00003• 5-10 repeats, 10 epochsIn my case I want my model looks like picture but with a little variation in the style and face.Knowing this I used the follow configuration settings for my model:AdaFactorSchedulerCosineUNET Rate: 0.0005Text Encoder (TE) Rate: 0.00003DR: 64 AR: 32So Here is my Parameter Settings on Tensor Art in-site training.First Upload your ImagesNote: Tensor Art Automatically tag your imagesNext select Profesional Mode in the settingsSelect the base model IllustriousNext click on NetworkModule menu and select Dora (better than LoRa)Write the trigger wordAnd then change the parameters to thisNext fill the positive and negative prompt of the image you want to generate as preview of your model while is training, and the image sise.Finally, click on Start Training NowWhen the training is done, here comes the hard part because you have to choose which one to publish on your profile or download.Usually is recommend the last one. So click on publish and click on create a projectChoose a A Project Name, Channel and Tags related to your modelIn the next window change the Version of your modelScroll down and change Showcases and Permission Settings if you want to.Finally click on Publish.And that's it.You can try it in your profile but you have to wait a few minutes until it’s deployed.This are the results for my model Here si the link if you want to try it!https://tensor.art/models/803009896502202836/Merry-Illustrious-or-Christmas-Walkthrough-FinalI Hope everything goes fine and you enjoy the results.Good luck and THANK YOU for reading this article!

Note: "I'm not a programming expert just a hobbyist who loves AI".Due to update Comfyui on Tensor Art some things has change a little bitIn this tutorial about Ai Tool we going to create a image to image Flux Model whit text description of what you want to change in the image you chooseAi tools are a great way to harness the power of AI especially to create or change the style of your image either to anime, comic, illustration etc. or change clothes like the one we going to create.You can create it to use it yourself or share it on the site for other users to use.Brief IntroductionBut to create an Ai Tool we must first start with the basics about them: The Workflow.I won't go into advanced concepts that complicate our lives; but I will tell you that the workflow is a set of steps such as loading a model, inputting prompts, and defining samplers, that the AI follows to create or transform an image or even a video.You can create a workflow or you can search for workflows on the site created by other users with the site's search tool.Creating a simple and functional workflowFirst of course you must have an active Tensor.art account.Second pass the mouse over Create and select ComfyFlowClick the New Workflow to enter the workflow interface Then Empty Workflow TemplateOnce inside we will double click on the workspace and a small window will appear to search for our first Node.A node is...this oneSelect TA Nodes and add TA Node LoadImage to the workflowNext double left-click on the workflow and search for VaeEncodeNext search for Loadvae Next search for samplercustomadvancednext search for vae decodenext search for save imageand conect them according to the next imageNext search for load diffusion modelNext search for dualcliploaderSearch for clip text encode (prompt)and finally Search for "Randomnoise", "Basicguider", "KSamplerselect" and "BasicScheduler"Here is the change made for Tensor ArtRight click on clip text encode and select "convert widget to input" and "convert text to input"Next click TA Nodes and select TA node prompttextand then connect them just like the follow imageand connect "Randomnoise", "Basicguider", "KSamplerselect" and "BasicScheduler" like thisFinally we upload our image, write the prompt that make some changes on the image and click on the generate button to test our workflow.We wait a moment and if everything goes well, we will see the final mixed generated image.Tip: Our workflow is automatically saved with every small change we make; so don't worry if for some reason it crashes unexpectedly or the pc shuts down.Note: We must always test our workflow and another workflow of the site to check if it is functional before publishing it because if we don't do this, Tensor.Art will not allow us to convert it into an Ai Tool.Click on the Publish button at the top left.Then a new window will appear in which we must choose whether we want to publish our workflow, only as a workflow or as an Ai Tool.Select Ai Tool and a new tab will open.Here the workflow will be selected automatically.We place the title we want to give to our Ai Tool.Then select the channel in the next drop-down menu, in our case, "Realistic".Select the relevant tags to our Ai Tool (Example: FLUX, INPAINTING).Then we must upload the final images that we obtained after having done the test in the workflow. If we download them we can select Upload or directly from workspaceIn the make contrast cover it is advisable to upload the initial image and the final image because a thumbnail is generated as a preview, so that if another user wants to use the Ai Tool knows what to do beforehand.Finally we click on publish and... ready we have our first Ai Tool, which we can share, if we want, in our social networks.And that's all folks.Enjoy it!

Note: "I'm not a programming expert just a hobbyist who loves AI".In this tutorial about Ai Tool we going to create a simple outfit change between two images choseen by ourselfsAi tools are a great way to harness the power of AI especially to create or change the style of your image either to anime, comic, illustration etc. or change clothes like the one we going to create.You can create it to use it yourself or share it on the site for other users to use.Brief IntroductionBut to create an Ai Tool we must first start with the basics about them: The Workflow.I won't go into advanced concepts that complicate our lives; but I will tell you that the workflow is a set of steps such as loading a model, inputting prompts, and defining samplers, that the AI follows to create or transform an image or even a video.You can create a workflow or you can search for workflows on the site created by other users with the site's search tool.Creating a simple and functional workflowFirst of course you must have an active Tensor.art account.Second click the Workflow on the sidebar of the siteClick the New Workflow to enter the workflow interfaceThen Empty Workflow TemplateOnce inside we will double click on the workspace and a small window will appear to search for our first Node.A node is...this oneSelect TA Nodes and add TA Node LoadImage to the workflowThen right clic on the node and Clone itNext search for LayerMask: SegmentAnythingUltra V2Next search for CatVTON WrapperAnd Finally search for PreviewimageNext we joint the nodes and set the parameters as followFinally we upload our images and click on the generate button to test our workflow.We wait a moment and if everything goes well, we will see the final mixed generated image.Tip: Our workflow is automatically saved with every small change we make; so don't worry if for some reason it crashes unexpectedly or the pc shuts down.Note: We must always test our workflow and another workflow of the site to check if it is functional before publishing it because if we don't do this, Tensor.Art will not allow us to convert it into an Ai Tool.Click on the Publish button at the top left.Then a new window will appear in which we must choose whether we want to publish our workflow, only as a workflow or as an Ai Tool.Select Ai Tool and a new tab will open.Here the workflow will be selected automatically.We place the title we want to give to our Ai Tool.Then select the channel in the next drop-down menu, in our case, "Realistic".Select the relevant tags to our Ai Tool (Example: FLUX, INPAINTING).In the text space we should put a brief description of what the Ai Tool does, what are the steps to be performed, what needs to be uploaded or written; what model, LoRa is used etc.Then we must upload the final images that we obtained after having done the test in the workflow. If we download them we can select Upload or directly from workspaceIn the make contrast cover it is advisable to upload the initial image and the final image because a thumbnail is generated as a preview, so that if another user wants to use the Ai Tool knows what to do beforehand.Finally we click on publish and... ready we have our first Ai Tool, which we can share, if we want, in our social networks.And that's all folks.Enjoy it!

Training a LoRa on any model could be challenging sometimes but in the case of Flux model is easier than SDXL or SD 1.5, especially if you have some guidelines.Here's my experience creating a "Twisty The Clown" LoRa on-site training feature.First thing firstTensor.art AccountSet up an account on Tensor.art and ensure it has a minimum of 300 credits.Dataset GuidelinesFor optimal results, it is advisable to include between 10 o 20 images in your dataset. The majority of these should be close-up photographs, complemented by a few mid-range shots and at least one full-body image. Additionally, incorporating side profile images alongside front-facing ones will enhance the dataset's diversity. To ensure efficient training and high-quality outcomes, the recommended image size is 512x512 pixelsHere is my dataset that I've used for this LoRa.Guidelines for Captioning Photos for LoRATensor.art automatically put captions on your images but consider adding:Unique Trigger Word: Begin each caption with a distinctive trigger word that identifies the dataset.Also consider this on your captions.1. Descriptive Tags: Include tags for all elements you wish to vary in the generated images. For instance, if you intend to modify attributes like hair color and length, ensure these are detailed in every photo caption. 2. Consistent Elements: Any aspect not explicitly described and that remains consistent across all photos in the dataset will automatically be associated with the trigger wordParameter Settings and Likeness CaptureUtilizing a collection of approximately 12 portrait photographs, a highly effective and adaptable likeness LoRA can be developed. The training employs default Flux settings, with modifications to enhance the accuracy.Here is my Parameter SettingsFirst select Profesional Mode in the settingsSelect the model FluxAnd then change the parameters to thisFinally, here are the results!The model link if you want to try it or downloadhttps://tensor.art/models/786414177230229338?source_id=njy_rlzlnUKzpfYra33z-xQhTHANK YOU!