Prompt Tutorial - Velma fom Scooby Doo (Patreon)

Published:

2023-01-21 20:21:12

Imported:

Content

I am going through my Raw Prompt Encyclopedia to clean up and improve prompts for you guys to use without much effort, and I thought this would be a great opportunity to show you my process to getting great generations with the Bastard line of models. Lets get right to it!

The Prompt:

```
realistic photo, samdoesart,masterpiece, (velma scooby−doo:1.2) highest quality, high quality, cute, detailed face+eyes, detailed face and eyes, glasses, nipple, short, looking at viewer, large breasts, (orange turtleneck sweater:1.2), shirt-lift, exposed breasts, livingroom, 18 year old girl, highres, (by Jeremy Lipking and Antonio J. Manzanedo:1.1), (by artgerm:1.1), open mouth, plump lips
```

First of all, we have potential for good generation here because this is a popular character throughout the decades, with multiple quality sources of data we can pull from. What I want to see is a movie rendition of the character, so I am going to try to pull movie-based prompt words to generate something good.

I am going to make the prompt use my experimental exploratory words first.

```
(velma scooby−doo:1.2)
velma_dace_dinkley
velma \(character\) \(cosplay\)
velma \(character\)
velma_dace_dinkley \(character\)
velma dace dinkley \(character\) \(scooby doo movie\)
velma
```

Unsurprisingly, I think the term `velma` is strong enough that we can just attach our prompt to that and it will carry through. So lets edit our prompt to close in on our target.

```
(velma), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot,
```

This gives me more specific details on my generation and gives more lighting. Reduces random generation by a lot. Lets continue to add:

```

(velma), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d

```

This gives me a computer digital art feel, almost video game character like. For the moment, I am going to keep it, but I might go a different direction. Next lets add in the artists from the prompt we are working from:

```

(velma), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d, (by Jeremy Lipking and Antonio J. Manzanedo:1.1), (by artgerm:1.1),

```

Its showing me Velma, but pretty stylized! I wasn't going for an alternate-universe stylized look, but these are not bad. Let's keep them for the moment while we work on clarity through the negative prompt.

```

Negative Prompt:

(blurry:1.1), lowres, low detail, bad anatomy, disfigured, poorly drawn face, ugly, cartoon, anime, illustration, multipanel,

```

This gives us a much better image, and it is pretty good on most fronts, background, foreground, character focus, pose, costume. Its just still blurry on the face and some elements. We will revisit this issue in a moment. Lets add some character details so we get more of what we are actually looking for! Lets get some more elements from the original prompt:

```

(orange turtleneck sweater:1.2), shirt-lift, exposed breasts, 18 year old girl

```

And upgrade it a little to this:

```

(orange turtleneck sweater, self shirt lift, micro bikini top:1.2), exposed breasts, 18 year old girl

```

And just put it at the end of our prompt:

```

(velma), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d, (by Jeremy Lipking and Antonio J. Manzanedo:1.1), (by artgerm:1.1), (orange turtleneck sweater, shirt lift, micro bikini top:1.2), exposed breasts, 18 year old girl

```

I want that action prompt keyword, self shirt lift, to show me something, so I am going to put it up front.

```

(shirt lift:1.1), (velma), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d, (by Jeremy Lipking and Antonio J. Manzanedo:1.1), (by artgerm:1.1), (orange turtleneck sweater, micro bikini top:1.2), exposed breasts, 18 year old girl

```

These are good, but its missing 2 key elements that I want to see together, orange sweater shirt lift, and bikini underneath. Put those up front:

```

(shirt lift:1.1), (velma), (orange turtleneck sweater, micro bikini top:1.2), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d, (by Jeremy Lipking and Antonio J. Manzanedo:1.1), (by artgerm:1.1), exposed breasts, 18 year old girl

```

This is really close to what I want to see! I lost a lot of elements in the generation, like the background is no longer coming through and the face clarity is not 100%. Lets try to tackle the clarity by increasing the resolution and turning on face restoration and hires fix. We are on 512x512 now, lets up it to 768x768:

That really enhances so many details. Lets do something fun with our character, give her a body expression and pose, and mess with some lighting and setting. For this we are going to split out our prompt into a second 75 chunk, starting off with our first 75 chunk as our template:

```

(orgasm, pleasure face, afterglow eyes:1.2), glasses, duck lips, dick sucking lips, ahego face, dynamic pose, porn lighting, bedroom, watermark, hires, 8k, masterpiece, highest quality, (velma:1.1) 18 year old young daughter, small breasts, nip slip, orange sweater, black bikini,

```

That gives our character much more depth. I was going for a lusty look and I think it comes out. Both the first prompt and this prompt is under 70 tokens, lets bump it up to 80 with commas. Then we have our pieces, lets combine them into a layered cake-like prompt:

```

Wow, I like the generation directions its going. This is good stuff.

Now I want to see Velma fuck. Straight up explicit sex in my face. Lets adjust our descriptors in both of our prompts:

```

(velma 18 year old girl having vaginal sex with a large man:1.2), (in pose together man behind girl girl on top:1.1), (orange turtleneck sweater shirt lift:1), watermark, hires, 8k, masterpiece, highest quality, boots, full body shot, unity blender unreal daz3d, (by Jeremy Lipking and Antonio J. Manzanedo:1), (by artgerm:1), 18 year old girl, , ,

(orgasm, pleasure face, afterglow eyes:1), (velma 18 year old girl having vaginal sex with a large man:1.2), (in pose together man behind girl girl on top:1.1), huge penis, vagina, penis in vagina, glasses, ahego face, dynamic pose, porn lighting, bedroom, watermark, hires, 8k, masterpiece, highest quality, , , , , ,

```

Sexes up in the direction that I want for sure. Lets add some bulk to our negative prompt just to see the generations:

```

(blurry:1.1), lowres, low detail, bad anatomy, disfigured, poorly drawn face, ugly, cartoon, anime, illustration, multipanel,

(((deformed))), [blurry], bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), messy drawing, malformed hands, malformed face, out of frame, out_of_frame, crop, preview, thumbnail, sample, mutation, mutated, closeup, (extra_limb), (ugly), animal, text, ((((bad anatomy)))), ((((deformed)))), ((poorly drawn face)), ((poorly drawn hands)), ((ugly)), (((deformed))), [blurry], bad anatomy, disfigured, poorly drawn face, mutation, (((poorly drawn face))), ((poorly drawn hands)), (disfigured)), long neck, ((two heads)), two faces, malformed hands, malformed face, (((mutation))), (((mutated))), [blurry], bad anatomy, (extra limb), (messy drawing), ((ugly)), (((deformed))), [blurry], bad anatomy, disfigured, poorly drawn face, mutation, mutated, (((pubic hair))), (((looking at the viewer))) fur, body hair, furry, breasts, (bad_prompt_v2:0.7)

lowres, (((bad anatomy)))), bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, breasts, ((large breasts)), woman, (((holding baby))), shota, shotacon, wolfgirl, dog girl, anime, drawing, illustration, (((body hair))), thick thighs, (ugly), fat, obese, (((deformed))), [blurry], disfigured, poorly drawn face, mutation, furry, mutated, (extra_limb), (poorly drawn hands), messy drawing, mutated, deformed, hairy pussy, extra legs, extra arms, bad body, disfigured, mutation, malformed, extra legs, extra arms, ((black and white)), (film noir), (greyscale), clothes, clothed, ((((portrait)))), ((((profile)))), landscape, environment, (((out of frame,))), anime eyes, large eyes, sunglasses, (bad_prompt_v2:0.7)

```

I am going to run this prompt as is through LiveAction and Premium to see what comes up:

The prompt words very similarly across the 3 models, so you don't have to do much rework at all! Amazing! You straight up get better everything in the generations as you go up, but a side effect is on Premium, Velma looks a little less like Velma, that bob haircut disappears a bit.

One final thing I will try is convert this through Premium i2i model to see our maximum lift from Anime/LiveAction/Premium:

Here is a few more end results of the Premium workflow:

There is a Part 2 of this tutorial that is pay-walled to Premium patrons only, and it is by REQUEST ONLY. Hit me up on Discord or Patreon for the details.