CivArchive
    Under The Rain - v1.0
    NSFW
    Preview 7826929
    Preview 7826934
    Preview 7826921
    Preview 7827134
    Preview 7826931
    Preview 7826926
    Preview 7826935
    Preview 7826933
    Preview 7826928
    Preview 7826916
    Preview 7827292
    Preview 7826919
    Preview 7826938
    Preview 7826918
    Preview 7827231
    Preview 7827255
    Preview 7826924
    Preview 7826915
    Preview 7826917

    Hi again and welcome to Under The Rain LORA.

    "Under The Rain" is part of my personal weather/ambient series where I aim to create an specific weather/ambient without using any specific word related to the weather or ambient condition. There is also some cool enhancements like in this case for example, inner Depth of Field (DOF).

    "Under The Rain" aims to create image/scene generations in a rainy weathery, wet surfaces, waterdrops, etc. These scenes will be influenced also by the Depth of Field also trained on purpose there to get a bit more realistic and less "plastic".

    Main token is t1hal and class is style so you can use t1hal style or just t1hal. I recommend the first option.

    This LORA doesn't have powered subtokens but there is a tendency when you use word "streets" to get more fidelity on some situations so mind this.

    You can use negative embeddings and words but this LORA has been proved to work without them in a normal way. Choise will be yours.

    How to use this LORA?
    *************************

    You can refer to the sample images I upload in the post but:

    - For characters * t1hal style Portrait/side portrait of concept... / a t1hal style concept...

    - For landscape and objects a [concept/object/whateveryouwnt] in the style of t1hal

    Which are the best values?
    *******************************

    - Extremely recommended 1 but you can use values from 0.8 onwards. You can also go higher for deeper impact but value 1 by default should work on most cases.
    - CFG Scale in this case is a bit different than usual. CFG Scale recommended is 2-2.5/3. You can also use 5 or 6 and onwards but it's not recommended. CFG Scale at 6-7 lose a big chunk of the inner aesthetics.
    - Clip Skip: At 1 the composition and dynamics are more variable and at 2 is more "standard". Feel free to try.

    Which models are good for this LORA?
    ********************************************

    - Tested on: Colorful, Rev Animated, Dreamshaper, Absolute Reality, etc.

    This LORA is GOOD for:
    ******************************

    - Character illustrations with weathery impact.
    - Depth of Fields compositions.
    - Habitual and common situations with a touch of realism and photo.
    - Photograph aesthetic creations.
    - Wet hairs and skins.
    - Wet surfaces and armors/suits, etc.
    - Dropplets
    - Rainy Weather on compositions.
    - Blue-darkeish colors aesthetics with DOF impact.
    - Standoffs

    IMPORTANT!

    The XL version was released for all our Patreon supporters as a way to thank them for the continuity support. If you want it, join us and this way you will also be helping me making more quality stuff for CivitAI.

    I wrote an article here and it's like my finetunning diary for the ones interested with releases upcoming, trainings done, exclusive patreon models previews, personal thoughts, etc. So if you want to see what I'm planning or what I'm into, feel free to take a look.

    As always thanks to everyone reviewing, giving feedback , commenting and supportin us in Patreon. Keep supporting us so we can keep creating and delivering quality content. Please support the Patreon since it's making this things to be possible around here.

    David

    Description

    V1.3 revised 3rd training for Stable Diffusion 1.5

    FAQ

    Comments (8)

    FrozenGunnerMar 13, 2024· 2 reactions
    CivitAI

    Following your instructions it works brilliantly. Thank you for this ❤️

    LDWorksDavid
    Author
    Mar 13, 2024

    No problem! Glad you liked it!

    olivettyMar 14, 2024· 3 reactions
    CivitAI

    Nice to see a new lora and what a lora it is! :D Can't wait to try it out, thanks so much!!

    FrozenGunnerMar 14, 2024· 2 reactions

    It can be a inflexible with prompts but It does a very good job for what it is designed to do.

    LDWorksDavid
    Author
    Mar 14, 2024

    Thanks Oliver! Sure, tell me how it goes 😄

    LDWorksDavid
    Author
    Mar 14, 2024

    @FrozenGunner yup. Happens if the concept is fighting a lot against the original style intention.

    FrozenGunnerMar 14, 2024

    @LDWorksDavid  Is that also due to the low CFG scaling?

    LDWorksDavid
    Author
    Mar 14, 2024

    @FrozenGunner I'm not sure... I have heard a lot that the more CFG scale the more you approach to the concept (and the guidance said the same even on earlier DIscoDiffusion era, 2 years and half back) but to be fair... I kinda have my own opinion on this thing and I think that a lot of people are just throwing random CFG scale (7-7.5 by def.) after training and that's a mistake (IMO, I could be wrong). I wrote about that here

    LORA
    SD 1.5

    Details

    Downloads
    1,396
    Platform
    CivitAI
    Platform Status
    Available
    Created
    3/12/2024
    Updated
    4/30/2026
    Deleted
    -
    Trigger Words:
    t1hal

    Files

    t1hal.safetensors

    Mirrors

    CivitAI (1 mirrors)
    Other Platforms (TensorArt, SeaArt, etc.) (1 mirrors)

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.