What you can do instead is ask the bot to specifically edit the setting field you want to edit. Chloe is a good start, but just copy the instructions over the input box and paste it into your chat message and tell It you want it to better write its own personality or whatever based on the guide pasted. It will do that properly if its given good instructions or if you add a bot guide rule in the persona box.

You have to make sure your prompts are concise and not too wordy. Leave out generic terms like masterpiece, bad quality, and go for technical camera prompts or photorealistic image of.... the brackets may not be the reason for the visible difference with flux. Having a good prompt structure is what works. Use Claude or chargpt to give better tips on how to setup your specific generator. Those have really made a difference in ensuring a more natural language flow in the prompting. I will test out the brackets too and see if that does help or not. But right now a guidance scale of 7 with a step of 8 for certain photoreal artstyles is working really good for me.

Here is some good photo prompts I use. I have the [input.artStyle.a] tag at the beginning of my prompt setup that then feeds into the character build for anime/fictional characters then into other prompts and have the [input.artStyle.b] tag towards the end of the prompt structure. This way flux see the key details of what to generate then sees the aspect of the artstyle. My biggest issue is with webcam, selfie, phone type cameras. Cannot get it to consistently generate real depiction. Only illustration or cartoonish images. I am still tweaking these as I go too.

  // Amateur
    ❗️❗️ Amateur
      a = snapshot photo of
      b = casual amateur photograph taken on a basic consumer camera, natural available light with no professional setup, honest unpolished framing, relaxed snapshot aesthetic with slight exposure imperfection.

    Cheap Snapshot
      a = early 2000s point-and-shoot digicam photo of
      b = shot on a Casio Exilim compact digicam, low-megapixel CCD sensor rendering with flat plastic-lens perspective, blown on-camera flash highlights, washed-out colors with crushed shadow detail, Y2K digicam color rendering with visible JPEG compression.

    Disposable Camera
      a = lo-fi disposable camera photo of
      b = shot on a single-use 35mm disposable camera, harsh direct built-in flash with unpredictable exposure, heavy silver halide grain, color shifts from cheap film emulsion, light leak streaks along frame edges, amateur spontaneous framing with physical film scratches.

    Fujifilm QuickSnap
      a = Fujifilm QuickSnap disposable camera photo of
      b = shot on a Fujifilm QuickSnap loaded with Fujicolor 400 film, cool cyan-shifted shadows with heavy silver halide grain, strong direct on-camera flash with deep crushed blacks and overexposed skin, 1990s amateur snapshot aesthetic.

    Kodak Gold 200
      a = Kodak Gold 200 disposable camera photo of
      b = shot on a 35mm disposable camera loaded with Kodak Gold 200 film stock, warm yellow-amber color cast with boosted reds and warm skin tones, medium silver halide grain, harsh direct on-camera flash with slightly overexposed midtones, 1990s family photo warmth.

    Pentax K1000
      a = amateur 35mm SLR film photograph of
      b = shot on a Pentax K1000 fully manual SLR with a 50mm f/2 lens, honest ungraded natural color reproduction with deep shadow rendering and visible silver grain, tactile film texture with accidental dust specks and a mild light leak from aging door seals, sincere student photography rawness.

    Polaroid
      a = instant Polaroid photograph of
      b = shot on a Polaroid 600 instant camera, square format with heavy white border, slow chemical development producing warm faded whites and milky midtones, slightly underexposed and soft rendering with dreamy uneven color development and barrel distortion from the plastic lens.

    Slumber Party Snap
      a = 2000s snapshot photo of
      b = shot on a cheap point-and-shoot at an indoor party, heavy lo-fi grain with light-leak streaks along frame edges, soft-focus nostalgia with underexposed sensor noise, muted color speckling and crushed shadow noise, candid and spontaneous framing.

  // Cinematic

    ❗️❗️ Cinematic
      a = UHD cinematic film still of
      b = professional color graded image with warm skin tone highlights and cool cyan shadow tones, complementary split-toned color grading, high-production commercial aesthetic.

    Blackmagic
      a = indie film still shot on Blackmagic Pocket Cinema 6K of
      b = shot on Blackmagic Pocket Cinema 6K with Blackmagic RAW color science, Super 35 sensor rendering with rich organic skin tones and 13-stop dynamic range, expressive moody shadow detail with shallow cinematic depth of field through an EF cinema lens.

    Cinema
      a = high-budget Hollywood cinematic film still of
      b = shot on anamorphic cinema glass in CinemaScope aspect ratio, deep professional color grade with selective shallow focus and controlled bokeh, rich textured shadow detail with subtle film grain, dramatic motivated practical lighting and epic production value.

    Cinema Blue Ultra
      a = experimental 35mm film photograph of
      b = shot on Film Photography Project Blue Ultra ISO 3 emulsion, dominant blue-violet tint throughout with popping saturated reds against muted neutrals, ultra-fine grain structure with no anti-halation layer, otherworldly alien tonal palette.

    CineStill Commercial
      a = television commercial 35mm film still of
      b = shot on CineStill 50D emulsion, ISO 50 daylight-balanced cinema stock built on Kodak Vision3 film base, vivid accurate color rendition with ultra-fine grain and broad highlight latitude, characteristic red halation glow around practical light sources.

    Fujifilm GFX 100 II
      a = medium-format cinematic portrait shot on Fujifilm GFX 100 II of
      b = shot on a Fujifilm GFX 100 II with 102MP large-format sensor, extreme subject-to-background compression with razor micro-detail at the focal plane, feathered ultra-shallow depth of field falloff with rich Fujifilm color science and studio-grade tonal gradation.

    IMAX Large Format
      a = IMAX 15-perf 70mm film frame of
      b = captured on IMAX 15-perf 70mm horizontal gate with massive frame area ten times standard 35mm, extraordinary spatial depth and immersive scale with near-grain-free image clarity, precise edge-to-edge sharpness in full 1.43:1 aspect ratio.

    Kodak Vision3 500T
      a = motion picture film still of
      b = shot on Kodak Vision3 500T tungsten-balanced emulsion, 3200K color science with cool blue-shifted daylight rendering, fine grain structure with clean shadow detail and broad highlight latitude, rich tonal gradation with a cinematic motion picture film look.

    Macro Beauty
      a = extreme close-up macro beauty photograph of
      b = shot on a Canon 100mm macro lens at f/2.8, ultra-shallow depth of field with clinical razor sharpness at the focal point, visible skin pores and fine surface texture with smooth creamy background separation, professional high-end beauty photography lighting.

    National Geographic
      a = award-winning National Geographic documentary photograph of
      b = shot on a telephoto prime lens with professional photojournalism composition, natural ambient color accuracy with subtle film grain, authentic on-location documentary photography gravity and moment.

    Nikon F
      a = 1960s photojournalism still shot on Nikon F of
      b = shot on a Nikon F SLR body with a fast Nikkor prime lens, sharp photojournalistic focus with high-contrast rendering, deep shadow density with tactile silver halide grain, raw honest press-photography aesthetic.

    Olympus Street
      a = 1960s half-frame street photograph shot on Olympus Pen F of
      b = shot on an Olympus Pen F half-frame 35mm SLR with a Zuiko 38mm f/1.8 lens, 18x24mm vertical frame with elevated grain from the small negative area, compact intimate street photography feel with 1960s Japanese documentary aesthetic.

    Panasonic Lumix S1H
      a = Panasonic Lumix S1H cinematic still of
      b = shot on a Panasonic Lumix S1H with V-Log L flat color profile graded for cinema, anamorphic lens desqueezed framing with soft digital highlight roll-off, warm controlled cinematic color grade through cinema-grade full-frame sensor rendering.

Try prompting it as - adult male, shaved smooth skin texture, no body hair. - Flux is receptive of a direct simole prompt and not using weights or SD type prompting.