ArtemSosyrev

With the release of Chester's character LoRA, I'm making changes again: since I know more or less about the trainer's condition and it's not going to be fixed, I'll upload/create LoRA again at any time, but at least every other day (occasionally two days in a row, one for each).

First, I feel a little better, so I can get back to making requests (and yes, I will try to transfer part of LoRA XL to PixAI, so don't worry those who wait a long time).Secondly, most likely, in the future, for brawlers skins from Brawl Stars, I will try to collect not 50 images, but more, so that the result is better. This change may also affect existing LoRA skins.And a question for you: in connection with the appearance of a character with two different models in the person of Kaze, I want to ask you what to do:Release two different versions for each variation (it may take a long time for both + need to choose which form is better to do first).Upload both variations in one version (faster, but there may be problems with the generation).Write in the comments how best to proceed (if the first case, then write in which variation you want to see it first).

You may all already know, but my most beloved Vtuber, Gawr Gura, recently announced her retirement from Hololive. To be honest, it was a low blow for me. After all, Gura was the first Vtuber that I started watching and the only one who always pleased me with her content. Although it was predictable to some extent, it still causes sad emotions.I can only wish my beloved baby shark good luck in the future. It is possible that sooner or later we will meet again.Since this news really hurt me, I need time to come to my senses. So, for now, without fulfilling your requests.Art from preview made by Amashiro Natsuki (one of Gura's designers).

Hollowstrawberry has messed up something in the code again, which is why it is now impossible to create a database for LoRA. So... Yes, again, without LoRA yet. Well, almost. I managed to create a base for LoRA with the character Edgar (Brawl Stars), so you will see one LoRA soon.Changed: not on the first of April, but I deceived you. With T4 and using stage 2 and 3, it works (unfortunately). So I'm going to have to be as economical as possible. It's also not the changes in the code that are to blame (they didn't even exist at that time), but the changes in Colab (unscrupulous Google, likes to break things all the time).

For some time now, I'm taking a break from creating LoRA XL for some reason:- I have some other project that is not fully related to AI, which I want to finish;- I need to upload the standard and XL versions to other websites (PixAI and Civitai);- Due to the fact that there is little free space left in Google Drive (where the databases for LoRA are stored if you use the trainer from hollowstrawberry), I need to clean it. But so that nothing goes to waste, so to speak, I will create standard LoRA characters that have outdated versions/ there are no standard loras.I can't say for sure when I will return to LoRA XL, but I can say for sure that the first LoRA XL after the break will be skins for Max and Bibi.

I found a funny thing with my LoRA: if you increase the value of LoRA above 1, then the model will adjust more to the character's style and make fewer jambs. I can say right away that this does not work with all models, so experiment at your own risk.

Here are all the plans for the coming days (marked in green as already done; yellow - still in progress or not fully completed; blue - still in doubt):Gallade from Pokemon (unplanned);Lumi from Brawl Stars (only version 0.5 so far);Pop Melodie (Melodie skin from Brawl Stars);Carnaval Mandy (Mandy skin from Brawl Stars);Queen Juju (Juju skin from Brawl Stars);Anyone from Brawl Stars (maybe Max first)Plans may change, please keep this in mind!

When there are enough materials, you will see her. And maybe some of the skins, who knows)

I wanted to surprise you by creating LoRA, a character that I had been thinking of doing for a long time, but decided on it only now, but something went wrong and the LoRA XL Trainer, developed by hollwostrawberry, failed. I'm already making another LoRA (updating the standard version of the LoRA character from Skullgirls, Umbrella) through a trainer made by the same person (it works fine), but tomorrow, if everything works fine with the LoRA XL Trainer, I'll try to do what I originally wanted.And no, don't suggest that I create LoRA inside the site. After working with the features of creating LoRA on PixAI, I don't really trust them, sorry.

I decided to tell you a little bit about myself (to give out a "base", so to speak).Birthday: March 28, 2003Nationality: Russian (I often use a translator to communicate with foreigners, so I'm sorry if something is unclear)Country of residence: KazakhstanThe date from which I started creating LoRA: sometime in February 2023 (at PixAI)The average attitude towards real artists: normal (without throwing mud at them and other similar stuff; not taking into account people who are directly against AI)Attitude towards the hostility of artists and AI users: extremely negative (seriously, stop already!)Main themes for LoRA: Brawl Stars, Plants vs Zombies, PokemonThat's about it. You can ask questions in the comments, as well as suggest requests for future loras (offer only XL/Pony, use PixAI/CIvitai for standard ones).

Merry Christmas (and Happy New Year) to all of you! I wish you all well!And now for the announcement. After I put all the LoRA skins on Mandy from Brawl Stars, I will step away from XL/Pony for a while. I owe quite a lot of regular loras to people with PixAI and Civitai (I didn't post anything at all on the latter for a long time due to difficulties with the site) and I don't really want to let them down now (and no, for me you are equivalent). Of course, there will be situations when I create LoRA XL/Pony, which will appear here, but these are more likely to be exceptions, because it's easier and cheaper for me to make images for LoRA here (they cost a lot of credits on PixAI, which is why almost the entire daily supply is spent on two XL images there). So, don't miss me.By the way, there are still loras that I might post here (because I was afraid to post them here because you might think something wrong about me), so I'll probably have some content for you (and I also have ideas for future loras).

As I said already under one of the models, I plan to create and publish LoRA XL/Pony of Brawl Stars characters here. I haven't decided who to start with yet. I can start with Juju, because so far I don't have a standard version of her LoRA, or with someone who already has a basic LoRA (but do it, so far, without skins). If you have any suggestions, write in the comments.I'm also thinking about the "resurrection" of some LoRAs that were made using the old method long before XL appeared (using the Gelbooru stage in the hollowstrawberry trainer). Among such Loras: 4 characters from Skullgirls (Parasoul, Valentine, Umbrella and Marie) and 1 character from Pokemon (Eureka/Bonnie). Because they were made using the old method, respectively, I do not have the images/database that were used for LoRA. Do you think I should bother?

If the clothes (or something like that) do not want to be displayed the way you wanted, then use (<clothing_words>:1.2). You can set the values higher, but not 1.5 or higher, because TensorArt does not support large values;LoRA with characters who are minors according to the canon will be marked in the descriptions. That is, they can be used for NSFW, but I do not recommend doing this (publicly, at least) to cause problems for yourself or me. So you are free to do it;Due to the strangeness of the mechanics of prompt words on the site, many words may not be displayed directly in the generator. Therefore, see the description of the version/LoRA to find the entire product (but be careful, the generator may break if you open another page of the site and have to restart it);LoRA can be released several times a day, or not for weeks. Everything will depend on me, my desire, time and amount of resources, because I create through a trainer developed by hollowstrawberry;ONLY LoRA XL will be here, look for the standard ones on my Civitai and Pixai accounts