Transforming 3D Characters into Realistic Portraits with AI - Stable Diffusion - Çift Dilli Altyazılar
Hey everyone, in this video, we're going to use a stable diffusion to transform ugly 3D people into amazing realistic characters.
Now, you might be wondering, why use real people in the renderings?
Well, sometimes 3D models offer more flexibility in terms of pose and lighting, but they can also look a bit plastic.
This is where a stable diffusion comes in.
This AI tool can take your existing 3D renders and add a layer of photorealism.
So how we do this?
First things first, you would need a stable diffusion set up on your computer.
If you don't have it installed yet, check out my tutorial in this link.
Now, the stable diffusion comes with models.
But let's just spend this.
We're going to download a new model called Cyber-Realistic from Civitei.
It's really good at creating realistic people, which is exactly what we need.
Head over to Civitei's website.
Find a super realistic model and click on the V4.1 back to basics.
and then the download button.
Once you have it,
go to the AI folder, then go to the Stable Diffusion Web UI folder, open the models, and then open the Stable Diffusion folder.
This is the folder,
where we are going to add the models we want to use with the Stable Diffusion,
so cut the CBER realistic with just download it and paste it in this folder.
Once you are good to go, run a Stable Diffusion by going to the AI folder we created in the previous tutorials.
At the top of the screen, you can find the Stable Diffusion checkpoint.
Change these for the cyber realistic we just installed.
If you don't see it, click on the refresh button.
Get to the image to image staff and then select the image pane.
This where the magic happens.
Here, you can drag and drop the 3D render you want to enhance.
But this option also allows you to paint in a specific area, like the face.
So let's do that and use the brush to paint over the face and neck of one of the characters.
If you have multiple 3D people, it's better to do each one separately.
Once you have done this, scroll down a bit and change the in-paint area to only mask.
We're also going to adjust a few parameters.
We're going to start with the noise in the string.
Think about this as how the algorithm is going to affect the image.
At zero, nothing will change.
Another one, we will get a non-related image.
We need to find the perfect balance, so let's try 0.44 now.
Then, let's change the batch size.
This option lets you generate multiple images at once.
Set this to a high number, like 4 or 5, to get a bunch of variations in one go.
And the last setting we're going to change is the sampling steps.
This will improve the generated image, but the higher the number, the more time is going to take the finish.
So let's start with 40.
Now the fun part, crafting the prompt.
This is part.
This is This is my favorite part.
of instructions for a stable diffusion, telling it what kind of transformation we want.
We want to focus on making the person look great.
list.
So, here is an example prompt you can use for this person.
Asian, woman, face, pretty, brunette, for a real and soft smile.
You can adjust these based on your specific needs.
Maybe your character needs a softer lighting or a specific hairstyle.
The more details you provide, the better the results will be.
We also want to add some negative prompts.
Think of the negative prompts as a way of telling a stable diffusion what you don't want to see.
For example,
if you're trying to create a four-wheeler but keep getting cartoon-like results, you could add cartoon style or animation to your negative prompts.
In this case, we're also going to include Longneck and Naughty.
Now, with the prompt set and your in-painting area defined, click the Generate button.
Stable Diffusion will create image variations based on your instructions.
If you don't like any of the images, adjust your prompt and the parameters we discuss.
Remember, all this is about experimentation and trying different settings.
After the stable diffusion finishes generating the images, we can see four different options, each with some variations.
If you don't like any of these options, try a different prompt, parameter or even a different sampling method.
After you're happy with the first character in the in-paint area, click the erase button to remove the mask.
Then, paint on the face of the next character.
You may need to also adjust the prompt depending on the character features.
After clicking the generate button again, you're going to get four additional images.
Once you're happy with the results of all your 3D models, click on this button to open the folder with all the generated images.
Then bring both your original 3D render and your new AI generated face into Photoshop as separate layers.
Make sure your face image is the top layer.
Add a layer mask to your AI face layer.
This would let us select high and reveal different parts of the face and hair so we can do a seamless blending.
You can also use a soft brush on the layer mask in paint with black or white to reveal.
of the face.
As the AI phase uses the same lighting and settings as the original render,
we usually don't need to do any additional adjustments, but sometimes you may need to enhance the color contrast of the image.
There you have it, with the stable diffusion and a little creative prompting, you can take your 3D renders to the next level of realism.
Remember, the AI world is all about exploration and learning.
Don't be afraid to experiment with different prompts, muggles and settings.
Çeviri Dili
Seç
Daha Fazla Özelliği Açın
Trancy uzantısını yükleyerek, AI altyazılar, AI kelime tanımları, AI dilbilgisi analizi, AI konuşma vb. dahil olmak üzere daha fazla özelliği açabilirsiniz.
Büyük Video Platformlarıyla Uyumlu
Trancy, YouTube, Netflix, Udemy, Disney+, TED, edX, Kehan, Coursera gibi platformlar için çift dilli altyazı desteği sağlamakla kalmaz, aynı zamanda düzenli web sayfaları için AI kelime/cümle çevirisi, tam metin etkileşimli çeviri ve diğer özellikler sunar. Gerçek bir çok yönlü dil öğrenme asistanıdır.
Tüm Platform Tarayıcıları
Trancy, iOS Safari tarayıcı uzantısı dahil olmak üzere tüm platform tarayıcılarını destekler.
Çoklu Görüntüleme Modları
Tiyatro, okuma, karışık ve diğer görüntüleme modlarını destekleyerek kapsamlı bir çift dilli deneyim sunar.
Çoklu Uygulama Modları
Cümle dikte, sözlü değerlendirme, çoktan seçmeli, dikte ve diğer uygulama modlarını destekler.
AI Video Özeti
Videoları özetlemek ve hızlı bir şekilde temel içeriği kavramak için OpenAI kullanın.
AI Altyazılar
Sadece 3-5 dakikada doğru ve hızlı YouTube AI altyazıları oluşturun.
AI Kelime Tanımları
Altyazılardaki kelimelere dokunarak, AI destekli tanımlarıyla kelime anlamlarını öğrenin.
AI Dilbilgisi Analizi
Cümle dilbilgisini analiz ederek cümle anlamlarını hızlıca anlayın ve zor dilbilgisi konularını öğrenin.
Daha Fazla Web Özelliği
Çift dilli video altyazılarının yanı sıra, Trancy ayrıca web sayfaları için kelime çevirisi ve tam metin çevirisi sağlar.