<h1 id="heading-5766">Intro</h1>I made it just for fun and as experiment to build a model good for augmenting professional photographs. I am using Nikon camera with bunch of vintage lens. I expect to build an SD model which is able to produce moody, cinematic pictures with nice smooth bokeh and "analog style". Please note, that I don't plan to train this model on any hardcore nsfw. Don't expect / request it from "cinero" models ;) My preference is art, beauty and emotions.<h1 id="heading-2662">Some tips on Prompting</h1>Few examples:<ul><li>"[grayscale : [dimmed colors : vibrant color splashes : 16] : 8]" - I call it "temporal trick". What it does is just make your prompt depending on current step. With this prompts SD will use "grayscale" on steps 1..7. SD will use "dimmed colors" on steps 8..15. On further steps SD will use "vibrant color splashes". I believe that there is no strict limits on nesting level. What you can do with it? You can effectively reduce the number of tokens SD process on each step (reduce the length of active prompt). On the first steps there is no sens to specify fine details. You only need to specify the scenery roughly. On the later steps there is no sense to spend tokens on describing the composition and lighting (I suspect). So, in theory, with this trick and big number of steps you can keep your prompt short and have build very rich prompt at the same time. PS: this prompt above force the SD to draw the scene with very little colors and super vibrant segments (I showed grayscale images where a subject has few vibrant hair curls or clothes parts). Probably, you can reverse this effect by making whole picture colored with some parts made grayscale.</li><li>[Audrey Hepburn : Milla Jovovich : 16] - you can have fun with smooth transition from one face to another with XYZ plot script in Automatic1111. Also, this particular temporal trick with face / body helps my model to render most realistic and correct anatomy. I suspect you can also implement dynamic LoRa weighting with this trick. If LoRa don't have a trigger word you can just put the LoRa token like [ &lt;lora: ...:0.42&gt; : &lt;lora: ...:0.99&gt; : 16] or you can use multiple levels of nested "trigger words" from different loras.</li><li>"shot on %Brand Name% %Lens Mark Name% vintage lens" - if you find the vintage lens names which SD have in its memory, then you have a chance to improve an "analog style" of your picture. I used to use "Carl Zeiss Sonar", "Nokton", "Helios 44-2", but cannot confirm that each particular lens model gives unique effect. If you have you own list of confirmed lens models, then please consider to share it with community in comments to this model [%PICTURE OF LEELOO saying HELP%]</li><li>In near future I plan to build a training dataset with many images shot on beautiful vintage lenses to bring old-school photography soul into this model. I will use some unique trigger word for that or will use a "vintage lens" (not sure yet).</li><li>use "perfect anatomy", "anatomically correct body", "anatomically correct hand", "perfect hands", "anatomically correct fingers", "perfect limbs anatomy" and similar anatomical phrases to increase the chance to get correct anatomy.</li><li>Us words smooth bokeh, swirly bokeh, depth of field, smooth background to increase the separation of main subject and scenery.</li><li>Use "turbulent fog", "mist" and "haze" with "mystical lighting" to get nice atmospheric picture with super noticeable depth of scene. Also use "early morning" and "blue hour" phrases if you want to get cold morning vibes.</li><li>Use "scary face expression", "surprised expression", "inviting expression", "lustful face", etc to increase the chance to get noticeable emotions on face and visible "body language". It works, but not yet very noticeable.</li></ul><h1 id="heading-5767">Priorities of this model</h1><ul><li>Cinematic photo-realistic pictures of female character (sfw, softcore nsfw)</li><li>Natural body, skin texture, [to be improved] environment (dirt, dust, stuff on floor, retro furniture and devices)</li><li>Realistic optical / photo effects (smooth swirly bokeh, analog film grain, aberrations [in progress]) of vintage lenses (Carl Zeiss Sonar, Jupiter 37a, Helios 44-2)</li><li>[To be improved] Urbex, abandoned, decaying interiors, depressive vibes, dimmed colors, fog, mist, vapor</li></ul><h1 id="heading-2662"></h1><h1 id="heading-5768">How it was created</h1>It is based on few merges of <a target="_blank" rel="ugc" href="https://civarchive.com/models/8030?modelVersionId=101080">Analog Madness</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/2661?modelVersionId=15640">URPM</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/15003/cyberrealistic">Cyber Realistic</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/25694?modelVersionId=105035">epiCRealism</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/28059?modelVersionId=121557">ICBINP</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/50000/cine-diffusion">Cine Diffusion</a> with coefficients in 0.18..0.35.It was trained with two datasets of carefully selected art photos with similar features (cinematic mood, atmospheric, charming anatomy, soft core / ero, retro interiors, morning outdoors, etc.). Total number of images in datasets: 600-700. Trained as LoRa with 20 steps per image using Kohya_SS then merged with coefficient ~0.3 into Merge of mentioned Checkpoints. Better to use with my LoRa with the same name to amplify the effect. <h1 id="heading-5769">Further improvements</h1>By priority:<ul><li>[done] Fix / Improve hand and fingers generation</li><li>[in progress] Improve gloom, bokeh, chromatic aberrations, spherical aberrations, light leaks and old analog film features</li><li>Fix / Improve feet and toes generation</li><li>[in progress] Add more urbex, abandoned, vandalized interiors and lost / forgotten outdoor scenery (suggest me good datasets pls ;)</li><li>Fine tuning / improvements of eyes and anatomy</li></ul>Feedback appreciated...

CinEro_SD15

CHILI - because of sharp focus on main subject and photorealistic rendering of body details.MANGO - because of sweet, lovely, bokehlicious scene composition with outstanding separation of main subject.I tried to train it hard on artistic beautiful female anatomy with keeping it creative enough to render interesting intricate scenery with bokeh.FEEDBACKs are WELCOME!Prompts suggestions:<ul><li>"swirly bokeh, smooth bokeh, depth of field, analog film grain, vintage lens" - to amplify the separation of main subject and lifelike look of the picture</li><li>"realistic skin texture, natural skin, perfect anatomy, anatomically correct body, perfect body, anatomically correct fingers, perfect face anatomy, anatomically correct eyes, ideal round cornea" - to increase the chance to avoid anatomical defects and amplify the realism of human body</li><li>Use DPM++ 2S a Karras sample - it is more stable. Other samplers with Karras scheduler also good. UniPC shows itself unstable but 2x faster and sometimes also gives satisfying results. Euler a - not tested (should give performance comparable with UniPC, but UniPC usually better). DDIM - sometimes good in ADetailer or HighRes Fix. Heun also useful for HR Fix and ADetailer only.</li><li>Use HighRes Fix with Latent upscalers if you want to add much more details. Use Img2Img based upcalers like UltraSharp 4x if you want to minimize aliasing issues. I prefer Latent scalers with Antialiasing. Latent upscaling can inject absent details into low-res picture. I2I upscalers aren't smart enough to enrich the 1st pass img.</li><li>Use AfterDetailer. In most cases you can be satisfied by only one ADetailer pass with "mediapipe_face_mesh_eyes_only". Set bigger "Inpaint mask blure" (I use 24px) and "Inpaint only masked padding" (I use 128). Use 1280 resolution for this pass and ADetailer will give you outstanding face and eyes quality.</li><li>Use "temporal trick" [prompt a : prompt b : nn] to specify different Loras for different stages of rendering. Thi way you can use body related Loras on early steps and eyes / skin details related Loras on later steps.</li></ul>

Files with SHA256: 7b17aa984fb2ef2b3c24b9aacb6e2f37809ea69f88c8b922eb67796a37e7eef1