見出し画像

【StableDiffusion】同じ人物で違う服・ポーズにしたい！

Seika_USA

こんにちは。アメリカ・シリコンバレー在住のSeikaです。

現在、StableDiffusionを使って、同じ人物で違う服・ポーズにすることを試みています。
まだ研究中ですが、1回目の比較で作った画像を検証していきたいと思います！

（2024年4月9日公開）


こちらのサイトを参考に生成しました。

今回試した生成の流れ。

1. ControlNetのタブを開く。
2. 似せたい顔の写真をドラッグ&ドロップで画像の指定箇所に入れる。
3. 画像に「Enable」と「Pixel Perfect」を適用。
4. コントロールタイプとして「IP Adapter」を選択。
5. Preprocessorに「ip-adapter_face_plus」、右側の設定項目には「plusv2」を選択。
6. 使用するStable Diffusionのバージョン（例：SD1.5）を選択。
7. 画像生成のために「ジェネレイト」ボタンをクリック。

その後、
・ControlNetを調整。
・Loraを調整。
・基の画像の変更。
をしながら、比較していきます。

こちらの画像を使って、生成します。

１. Control Weightの比較

Control Weightを比較していきます

コントロールネットのウェイトは、生成される画像における特定の制御機能の影響力を調整するためのパラメータ。
ウェイトが高いと制御機能の影響が強く、
低いとテキストプロンプトなど他の要素が優先されます。
似たような画像を一貫して生成したい場合は、
コントロールネットのウェイトを高く設定することが有効。

では、まずWeight: 0。

プロンプト
masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket
Negative prompt: 16-token-negative-deliberate-neg,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
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 76735295, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plus_sd15 [d86a490f], Weight: 0, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Version: v1.8.0

テキスト（名作,最高品質,1ガール,美少女,上半身,東京の夜,ロングヘア,茶髪,ジャケット着用）が優先されていと思われます。
２枚目
確かにテキストプロンプトに忠実かも。

今度は、少しあげます。 Weight: 1でやってみます。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3204582840, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Version: v1.8.0

A Iっぽさが減った？
２枚目
写真に近づいたような、そうでないような

今度は、最大値Weight: 2まで上げてみます。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2085322147, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Version: v1.8.0

あんまり変わらない？肌が白い。
気持ちオリジナルに似ている？
どうでしょう・・・。

２. Loraを入れてControl Weightを比較

LoRAとは、あるモデルに対して追加学習を行ったファイルを意味します。LoRAを活用すると、生成する画像のクオリティを劇的に上げたり、
好きな画風や特徴を画像に加えたりすることができるようになります。

今度こちら↓のLoraを入れ最大値を１にし、Control Weightを比較していきます。

ip-adapter-faceid-plusv2_sdxl_lora

まずは低いウェイト、０で。テキストプロンプトが優先されるはず。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 402528609, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 0, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

雰囲気がグッと変わり美人系に。
同じ雰囲気の女性。
２枚目と似てる。

今度は、 Weight: 1。
テキストプロンプトは弱くなるのでしょうか。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2090606775, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

かっこいい女性になりました。
２枚目。基にした画像とちょっと似てる？
先ほどの２枚目の女の子と似ているかも。

Control Weightを最大値に。
オリジナルの顔と似せることはできるでしょうか。

こちらの女の子を基にします。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3493474794, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

・・・似てない。
アングルのためか、１枚目とはまた違う顔。
可愛いけれども。

３. Loraの強度を比較

先ほどの２. Loraを入れてControl Weightを比較ではLoraを最大値にしたので、今度は0.5→0.2と徐々に弱めていきます。

まずは半分の０.５。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket <lora:ip-adapter-faceid-plusv2_sd15_lora:0.5>

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 151795126, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plus_sd15 [d86a490f], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

美人系からかわいい系に
幼くなりました
目が大きくなりました。

では、もっと下げて0.2にします。

masterpiece,best quality,1girl,bishoujo,upper_body, tokyo night, long hair, brown hair, wear jacket <lora:ip-adapter-faceid-plusv2_sd15_lora:0.2>Negative prompt: 16-token-negative-deliberate-neg,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,blurrySteps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 151795126, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plus_sd15 [d86a490f], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

10代の子に見えます。
Loraを下げると幼くなるのでしょうか？
素朴な感じ
目が大きくなりました。

４. 似せたい顔の写真を変更

今度は、似せたい顔の写真を変更して比較します。

LoraとControl Weightは最大値にしてみます。

masterpiece,best quality,1girl,bishoujo,upper_body, natural Light, long hair, brown hair, wear whitedress <lora:ip-adapter-faceid-plusv2_sd15_lora:2>

Negative prompt: 16-token-negative-deliberate-neg,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

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1585127713, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

ん！？？
あれ？？
服をwear whitedressとしたのが失敗の原因？

今度は、AIで生成した女の子を基に生成。

先ほど生成した画像を基にします。

LoraとControl Weightの最大値はそのまま。
プロンプト、ネガティブプロンプトは変更。

Top Quality, 1 Beautiful Woman, Wearing White tanktop, white Denim pants, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3560723597, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

同一人物という感じはあまりしない・・・。

服、照明、表情を変えたらまた違うかも！？

Top quality, 1 beautiful woman, long Hair, brown hair, wearing Camisole & skirt, shy-smile, Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2687423748, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

爽やか〜。
オリジナルの子と似ている気がします。

服を変更。

Top quality, 1 beautiful woman, long Hair, brown hair, wearing Halter top& skirt, shy-smile, Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

先ほど生成した子と同一人物っぽい。

表情を変更。

Top quality, 1 beautiful woman, long Hair, brown hair, wearing Halter top& skirt, half-smile, Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

似てる。

ガッカリした顔ではどうでしょう。

Top quality, 1 beautiful woman, long Hair, brown hair, wearing Halter top& skirt, Disappointed, Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

似てるかも。

イライラすると・・・

Top quality, 1 beautiful woman, long Hair, brown hair, wearing Halter top& skirt, Irritated, Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

同じ人物が怒っているように見えます。

服をチェンジ。

Top quality, 1 beautiful woman, long Hair, brown hair, wearing casual dress, shy-smile Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

同じ子っぽい。

今度はスーツ。

Top quality, 1 beautiful woman, long Hair, brown hair, wearing suit, shy-smile Sunlight, at street, 35mm Lens, /1, (white Background: 1.3) <lora:ip-adapter-faceid-plusv2_sd15_lora:1>

Negative prompt: sfw ng:deepnegatives:v.1.7.5t, ((badhandv4)), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), watermark

Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3998585933, Size: 512x512, Model hash: 1a17bcd93d, Model: beautifulRealistic_v7, ControlNet 0: "Module: ip-adapter_face_id_plus, Model: ip-adapter-faceid-plusv2_sd15 [6e14fc1a], Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", ControlNet 1: "Module: ip-adapter_face_id_plus, Model: None, Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced, Hr Option: Both, Save Detected Map: True", Lora hashes: "ip-adapter-faceid-plusv2_sd15_lora: a95a0f4bdcb9", Version: v1.8.0

いいかも

もしかしたらLoraとControl Weightの最大値にして、AIで作った女の子を基にした方が、似せて作れるのかもしれません。
まだ答えは出ませんが、引き続き比較していきたいと思います！

もし記事が役に立ったと思ったら、ぜひスキボタンを押してくださいね😊
それでは、次回の記事でお会いしましょう！

Follow me here😊

Photo by Lau keith on Unsplash

いいなと思ったら応援しよう！

Seika_USA
よろしければサポートお願いいたします。いただいたサポートはクリエイターとしての活動費に使わせていただきます。

この記事が参加している募集

#AIとやってみた

47,684件