CivArchive
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined

    This series is focused on blending cinematic themes with SFW/NSFW. It can generate a range of film, fantasy, realistic, surreal art and fine art. It is great at HDR quality images and cinematic images, with maximum detail.

    If you're a fan of images that seem to pop out at you, then this model is for you. This series will always focus on cinematic aesthetics and NSFW ready.

    Prompt weight guide below.

    Tip Me

    Donations to my Whop account will exclusively go towards hardware upgrades. Civitai Buzz will be used for site generation and promotions.

    Upgrading my local hardware is my on-going goal so I can train locally.

    Local training is imperative today. With Ai data centers there will be a vast drain on resources in the areas they operate. Since model & lora training will no doubt be hosted in the new centers, I want to be able to run locally only. The moment my Ai hobby causes actual harm, on any scale, is when I start collecting stamps instead.

    General Prompt Guide

    Weights in my prompts are primarily for testing the model's capabilities and trying to uncover limitations during test generation. However, I've created the tutorial below to help understand how model weights can be used more effectively.

    Positive Attention

    • Weights between 0.5 & 2.0 are usually sufficient.

    • I sometimes use weights between 2.0 & 3.0 when I want to force attention to a specific ethnicity for testing.

    • Higher weights are also helpful when you want to magnify size or specific details. The difference between: large breasts, (large breasts:1.2), & (large breasts:1.8) is obvious.

    Recognized weights are usually -5.0 to 5.0 with Dynamic Prompts.

    Attention Equation

    The parentheses ( ) on their own adds attention weight of 1.1. Unless you specify a weight.

    • (hyper-realistic) = 1.1

    • ((hyper-realistic)) = 1.21

    • ((hyper-realistic:1.3)) = 1.43

    Brackets [ ] are negative attention weight of 0.91, not negative weight, it's just attention that starts under the 1.0 default of all words without weights. Stacking them decreases that number.

    • [deep shadows] = 0.91

    • [[deep shadows]] = 0.81

    • [[[deep shadows]]] = 0.72

    Negative Attention

    • Weights over 3.5 are not usually effective or are a detriment during tokenization.

    • Negative weights do not have much effect in prompts.

    • Using too many high weights in the same prompt can cause distortions & deformities.

    • Using high weights will give specific keywords more attention, but that is often at the expense of other keywords.

    Bonus

    Prompt with high weights:

    (HDR:2.2), cinematic, (natural sunlight:1.23), (hyper-realistic:1.45), (full body shot:0.6), masterpiece, incredible details, [feeling of solitude], mysterious, (deep shadows:0.72), Beautiful (30yo Malaysian woman:2.52), hazel hair, (mid_length grungy hairstyle :1.43), (natural large breasts:1.6), (wearing an unbuttoned stylish silk shirt), (brown pearl g-string, leather lace sandals:1.41), (sexy pose, reclining slightly on elbow:1.68) on a stunning designer wicker couch, at a rustic beach house rooftop with driftwood furniture, seashell decor, a view of the ocean, perfect for casual summer parties, sensual atmosphere, beautiful, guilty pleasure, shot on Canon EOS 5D, RAW.



    Prompt without high weights:

    (HDR:0.86), cinematic, (natural sunlight:0.8), (hyper-realistic), (full body shot:0.6), masterpiece, incredible details, feeling of solitude, mysterious, (deep shadows:0.72), Beautiful (30yo Malaysian woman:1.32), hazel hair, (mid_length grungy hairstyle), (natural large breasts), (wearing an unbuttoned stylish silk shirt), (brown pearl g-string, leather lace sandals), (sexy pose, reclining slightly on elbow:1.12) on a stunning designer wicker couch, at a rustic beach house rooftop with driftwood furniture, seashell decor, a view of the ocean, perfect for casual summer parties, sensual atmosphere, beautiful, guilty pleasure, shot on Canon EOS 5D, RAW.

    Description

    Settings I use:

    DPM++ SDE

    35 steps

    5.4 CFG

    FAQ

    Comments (8)

    zigzag123Apr 23, 2026· 2 reactions
    CivitAI

    Outstanding next level SDXL model, out of the box great gens, great work..thank you!!

    jrrytng467Apr 24, 2026· 3 reactions
    CivitAI

    oh nice, this is going to be fun tonight.

    AFD_0Apr 28, 2026· 2 reactions
    CivitAI

    Actual quick review. Intrepid XL v2.0 is crazy good,

    AFD_0May 14, 2026

    And.. v3.0 is excellent is well, but thinking I might like v2.0 a bit better for a few reasons. Image quality and realism seems mostly comparable between both, though v3 doesn't do as well with full-body or similar subject distances further away than close/mid-portrait, but oddly, it usually does better in that regard at closer shots vs v2.0 which is interesting. After using a wide variety of prompting styles and various settings, v2.0 seems to have better prompt and LoRA adherence, and is just all-around a lot easier to "dial-in" to get the results you want. But.. when you finally get everything working as expected with v3.0, the results are hand-down a lot better looking and more creative than with v2.0 - and then you slightly adjust something, everything changes and you need to re-tweak everything all over again, lol.

    Both are excellent and ime, v3.0 can produce better results with some luck and/or tweaking, while v2.0 produces results almost as nice, but far more often and without as much tinkering and experimentation. Think with more time and learning, I'd probably grow to like v3.0 more than v2.0, but think I'll be playing around with both for a while!

    AFD_0Jun 6, 2026· 1 reaction

    And.. wow, v4.0 is back to crazy good and then some! Absolutely amazing image quality, excellent prompt and LoRA adherence, and a refreshing variety of artistry that keeps things fun without taking too many liberties. Does facial expressions and abstract concepts very well. Very flexible and easy-to-use with a wide range of CFG, variety of steps and length of prompt.

    Think my only critique would be that I'd prefer a slightly richer, slightly more saturated/colorful image output with just a touch more contrast, but that's really just a subjective aesthetic/stylistic preference and nothing that couldn't be adjusted with a different VAE or maybe a few LoRAs (probably, haven't really tried yet, but realism-enhancing LoRAs stack on this model quite nicely). (Edit; just to update, such output is 110% possible).

    All-in-all, a vast improvement all around over v2.0 and v3.0 imo. @ironically_irascible Very excellent work and thanks for sharing this with us!

    AFD_0Jun 23, 2026

    @ironically_irascible And.. dang, you nailed it! Both Reality Bound Clover v15 and Intrepid v5.0 are both huge improvements all-around (over v14 and v4.0, respectively). Image quality is better for both, prompt and LoRA adherence is even further improved (especially with negative prompting, which is often ignored/muted with so many XL-based models), and realism seems better (much more so in RB, but still improved from Intrepid v14 to v15 and the improved adherence might be picking up various realism LoRAs I always stack at very low levels).

    For me, the biggest quality-of-life improvement for both has been ease-of-use and flexibility of a wider range of CFG, steps, LoRA strengths and prompt lengths (ie, getting excellent results no matter how I use it). And out-of-the-box, I'm now getting excellent, as-expected lighting, contrast and saturation for both, and if needed, dialing everything in doesn't require much tweaking of prompt/settings. Only slightly weird thing I've noticed is Intrepid v5.0 sometimes boosts [[less emphasis weighting]], where as RB v15 seems to handle that really well. Not really an issue at all and probably just something I'm using incorrectly, since I only started using square brackets since that model. ((positive weighting works great+)) in both the positive and negative prompt and aligns very nicely with my naturally grouped and (constantly_weighted) prompting style.

    Excellent work on both Intrepid v5.0 and Reality Bound v15. I really can't decide which of them I like better. They're both so enjoyable to use, wish I could 50/50 merge them. Great job and thanks for sharing!

    dezeptup125Apr 30, 2026
    CivitAI

    Could you please share the optimal settings (and resolutions if specific recommended)? The pic metadata seems to be corrupt on .red now.

    ironically_irascible
    Author
    May 1, 2026· 2 reactions

    I always add the settings I like to use in the "About Version" section of each model. For V2 I use these:

    DPM++ SDE

    35 steps

    5.4 CFG

    My default resolution is width 896 & height 1024- upscaled by 1.5 in Hi-res Fix. But these are just my own preferences. Some images can come out even better if I use Euler A karras as the first sampler and then switch to DPM++ SDE for Hi-res. About 25 steps for Hi-res.

    Checkpoint
    SDXL 1.0

    Details

    Downloads
    700
    Platform
    CivitAI
    Platform Status
    Available
    Created
    4/23/2026
    Updated
    8/20/2026
    Deleted
    -

    Files

    intrepidXL_intrepidV20.safetensors

    Mirrors