【AI生成教學】如何生成專業攝影照? - Stable Diffusion 教學

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
Hello everyone, I am the AI ​​summoner Ah Zheng. Today I will share with you an interesting topic. How to make the generated photos look like those taken by professional photographers. When generating real photos, you will find that AI is actually not that smart. Sometimes The pictures that come out look very ordinary and featureless, like the pictures taken by a stupid boyfriend, such as this one, but if you just change the prompt, the pictures will become more beautiful and more textured, like a professional photographer. If you want to know, then let’s continue reading. Now let’s demonstrate how to turn an ordinary photo or even a photo taken by a stupid boyfriend into a photo taken by a photographer. It’s the same as the photo, we prepare a photo here, it looks like the photo that ordinary people would take, ok, what’s the prompt for this photo? Look here is a fashionable woman standing in the park at night. His background is Taipei 101. She has a delicate face. Just leave the Lora behind this reminder because I need to make some of her face here. Correction, so I added it. This is not the scope of today’s discussion. I only put one negative reminder, so that some things that are not suitable for children will affect our demonstration later. In order to wait for the demonstration, I will also use the seed here. I will also use controlnet to fix the movement of the characters. We can compare it better. If there are more prompts and prompts, what will happen in the future? Well, let’s start to make it look like a photographer. If this is the case, is this kind of photo called a photographic work? Then we will add a photographic work prompt, phtography ok. Since it is photography, it is not a painting, so we don’t want paintings, and it is not a sketch. Well, if it looks much better. It’s just that the hand looks a little weird. It’s okay. This will be corrected later. How to make the hands and feet look normal? After this, we will open a special topic to mention it to everyone. Let’s not worry about it here. In short, you find that the quality of this photo has improved. Does it look better? This boyfriend has evolved to another level. Okay, let’s come again. If it’s a photography work, should we? Want a good photo op? So we must be a masterpiece with a very good quality and best quality. Most people think that this is almost the same, but ours is different. We have negative prompts. What should be added is a big mistake and don’t come to bad things. Don't come to worst quality, don't come, we can double it, besides worst quality, we have a lower quality, don't come, that's double, and mediocre works, don't come, don't come, this is Twice as good and then what do you think it's over? If not, we can add it. We can add it because it is a high-quality work, so we have a work with a resolution of 8k. Then we also have a very high-quality photo such as HDR. The level of the photo is actually In this way, when we write these reminders into AI, he thinks that he has learned what kind of characteristics he has when he sees these reminders, and then he will put these characteristics in this photo. So We will write all the characteristics of some of the better works, and then he will slowly improve the quality of this work. Let’s generate it to see if there are any of these prompts. After downloading, the texture will immediately improve several times. level, right? Take a look this way, it looks good, yeah, so is that the end of it? Not yet. What else can we add? You will find that this is at night. There is not enough light at night. We want to make the colors sharper, so how do we do it? We want to add these prompts. We need to add raw photo first and then highres to make it sharper. What else should we add? absurdres Okay, let’s add it to 1.2 so that the reverse prompt is here, we write lowres, which means low resolution, we don’t want us to take a look, we can compare the previous photo, and you will find that it seems to be brighter and the photosensitive effect is better. What else can we add next? In the field of portrait photography, there is a very useful film called Kodak Portra 400. If we add this reminder, this photo will have Kodak Portra. Let’s take a look at the Kodak Portra 400. If we use film to take pictures, we hope to restore some of the features seen on the film. For example, when the light is relatively dark, we will see some fine particles. Those grains are called film grain. Well, in this way, we can simulate the texture of the film. Let’s take a look. We found that after using these two prompts, the light on the skin of the character becomes much softer because this is Kodak Portra 400 It is a feature and if you zoom in carefully, the black area at the edge of the hand junction actually has some grains of grains like some of the features that the film used to have. Now that we have all taken portraits The characteristics of the film have been put in, and then we start to make some adjustments to the picture. I want to blur the background of the picture, so how do I do it? It’s very simple. Blurry background. Let’s take a look. You’ll find that the background is blurred and the characters will be clearer. The background sets off our characters and they look very beautiful. If it’s blurry again, there is a kind of light spot. A term for bokeh is called bokeh. How can it be called bokeh? Use the word bokeh, then this term is called bokeh. Let’s take a look at the effect now, like here, it ’s all bokeh effect, but it doesn’t look so obvious. Let’s strengthen it. It seems that there are changes here and here. Is there any way this looks better and the light spot is brighter? Let's add some flare. Lens flare looks like a flare. He has nothing to add here. Here he enlarges it. Here he enlarges it. But here There is an extra building. Sometimes we tell him that we want halos, so he will add other light sources. Sometimes when everyone is generating, he will not really add halos, but he will add other things. Come in and increase the light, let’s take a look again. If I want to generate a photo with more color, I will add this vibrant color, and the generated picture will be more colorful. If you want to add more, you will find this one. The color of the picture, lighting, etc. are more than the previous one, and the color appears. Well, this one is the last photo we only need to produce. Then our demonstration today is here. Let’s review what we just did. What did we do? First of all, we told AI that this is a photography work, so we wrote photography on the front prompt, and filled in paintings and sketches on the back prompt, saying that we don’t want paintings or sketches. Second, we told AI that we need works of excellent quality, so we wrote masterpiece best quality on the positive prompt, worst quality, low quality normal quality on the negative prompt, and made two copies of each. Do not let low-quality pictures disappear. Third, we told AI that we want high-resolution photos, so we added 8K, HDR RAW photo highres high-resolution absurdres ultra-high-resolution plus In addition to enhancing the resolution, these words are also used because high-resolution photos are usually taken by photographers, so many places are worth learning. AI will take advantage of them and put them into our works to increase the visibility of our works. Reverse For the reminder, we put lowres, which means low resolution. Fourth, we began to tell AI that our photos should have the characteristics of film, so we put the two reminders of Kodak Portra 400 and film grain. What is film grain? It is the kind of particles that are only found in film, like the picture on the left below. The two pictures below are to show the texture of film. You will find that this kind of picture is not as sharp and colorful as the color taken by digital cameras. The texture of old photos Fifthly, we told AI that we need to blur the background. After blurring, we need to add bokeh and halo. This kind of aperture is called bokeh. The sun behind the cat in the lower right corner is called halo. Sixth, we told AI that our work should be rich in color, so we added a vibrant color and emphasized it to 1.2. What is a vibrant color? Please refer to the picture below and you will find that each picture has a variety of colors. Finally, seventh, we have a reminder. If you want to produce the kind of pictures that are not suitable for children to watch with the whole family, you must write on the front. If you want to put family friendly on the opposite side, put nsfw. What does it mean? It’s called not safe for work. It’s not safe to watch while you’re at work. But you know, if you just want pictures that are not suitable for children, just use it in reverse. Don’t say it. I taught you well. Let’s look back here. Let’s enjoy the series of changes in all the pictures we just made. Thank If you like it, please help me like, subscribe and share If you have any questions, you can leave a message for me, and I will try my best to answer your questions. If you want to know any other topics, you can also leave a message below. I will make some generative AI topics that meet your needs. Thank you for watching and see you next time
Info
Channel: AI召喚師
Views: 136,863
Rating: undefined out of 5
Keywords: stable diffusion, ai生成, 真人, 專業攝影, 人像攝影, chilloutmix, prompt教學, 提詞教學
Id: 4FQ8Mlolr0g
Channel Id: undefined
Length: 13min 22sec (802 seconds)
Published: Sun Mar 19 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.