Volver a discusiones
Consejos

免费版的七个模型生成的实拍风图像对比

ねねむ
Traducido a español·

Do you think that "realistic images generated by AI all look the same, even if the model is different"? You probably do, right?

Well, actually, each model has its own distinct characteristics!

It's not surprising if you didn't know that. There's hardly any prior research on this site comparing the distinctive features of photorealistic images across models—only comparisons of mecha images so far 🤖 (see by yamabico).

So this time, I systematically investigated the characteristics of photorealistic images from seven models available for free on this site 📷️

The common prompt was:
"photorealistic style, park, trees, autumn leaves, black streetlamp, night, bench, starry sky"
and the seed values were also the same 🍁🌃

(Note: Since only 5 images can be attached per post, some posts contain two images combined.)

① Image V3
image
The contrast between light and dark is precise and clear, with the shapes of tree branches, leaves, and benches well-formed. It has outstanding realism, like a photograph! 📷️

② Anim4gine
image
The streetlamp is slightly glitchy, but the lighting is soft, and the depiction of tree leaves is detailed. It has the feel of a hand-painted picture! 🖌️

③ Animij & WAI
image
Animij renders tree leaves with soft gradients, and the contrast between the sky, dark trees, and autumn leaves is vivid, giving it a picture-book atmosphere! 📔

WAI's depiction of tree leaves is somewhat rough, but the texture of the road is detailed. The placement of the streetlamp is also well-balanced, with excellent compositional stability! ⚖️

④ Neta Yume
image
The streetlamp and bench glitched, but the shapes of the tree leaves are crisp, and the use of yellow and orange colors is vibrant. The starry sky is also realistic, and the overall picture looks like a scene from an anime! 🎥

⑤ Image V2 & SDXL
image
Image V2 features very neat tree leaf shapes and streetlamp decorations, with delicate color usage creating fine gradients. It has astonishing detail precision! 🔍

SDXL shows some instability in rendering streetlamps and branches, but the bokeh of light, the richness of the background, and the shading of tree leaves are excellent. The overall effect is a fantastical atmosphere! 🔮

[Summary]

📷️ Image V3 excels in realism,
🖌️ Anim4gine in its painterly quality,
📔 Animij in color contrast,
⚖️ WAI in compositional stability,
🎥 Neta Yume in its anime-like quality,
🔍️ Image V2 in detail precision,
🔮 SDXL in richness of information.

This is just one example, and changing the conditions might reveal different characteristics for each model, so it's worth trying various settings 👍

Finally, I personally prefer Image V3 for its photographic atmosphere and Image V2 for its wonderful fine details—which one do you prefer?

①Imagen V3Votar
②Anim4gineVotar
③ AnimijVotar
④WAIVotar
⑤Neta SueñoVotar
⑥ Imagen V2Votar
⑦SDXLVotar
Vota para ver los resultados·Selecciona todas las que correspondan·Inicia sesión para votar
3

4 respuestas

こ

面白い考察だと思います。参考にさせて頂きます。どうも有り難う御座います😀👍✨🕊️

ね
ねむOP·

こちらこそです!
素敵な作品を待っています😊

ま
まれん2·

少し歪んだSDXLが、リアルなのにどこか夢みたいな雰囲気でとても良い👍🍁
Animijはベンチが洋風でおしゃれ!

ね
ねむOP·

確かに、SDXLのは少し歪んでいるからこそ夢のような雰囲気を演出できているのかもしれませんね✨
Animijのベンチは手すりと背もたれに曲線美がありますよね!

Dejar una respuesta

Únete a la comunidad para compartir tus ideas y pensamientos.

Inicia sesión para participar en la discusión