Krea AI-AI video generation

A brand-new video generation tool that generates videos through keyframes and text prompts, and users can flexibly move and adjust them on the timeline.

Super powerful instant image generation website

Although there are many AI drawing tools now, the most common one we use is “Venture Drawing”. At most, we just throw out an original picture to draw the drawing, and usually let the AI ​​start the calculation after giving the instructions. The biggest feature of the picture generation website Krea AI that I will introduce this time is that it can be generated while the user is drawing pictures or giving prompt words. With each command, you can see the changes in the pictures, and the picture produces pictures and texts. The two functions can be used together. When using sketches to draw pictures, if the drawing skills are not good or the AI ​​​​does not recognize the parts, you can use text to supplement the explanation. Of course, Krea AI does not only have the function of generating images in real time. Like many AI services, it has paid and free versions, so this article will introduce it in detail, and will also teach the three main functions later.

What are the advantages of instant generation?

Common AI drawings such as Midjourney, Stable Diffusion, etc. are all based on the LDM diffusion model (Latent Diffusion Models) algorithm. When generating a picture, we will see that a picture gradually becomes clear from blur. This is because it needs to go through ” “Removal of noise” process, and each step of noise reduction takes time, so it is relatively time-consuming. Krea AI can achieve “instant feedback” by using the technology of LCM (Latent Consistency Models). Its characteristic is that it only takes 2 to 4 steps to synthesize high-quality images, which can actually be achieved in “seconds”. “Plot”, which greatly saves the waiting time for generation.

If you use the LDM model algorithm to generate, you often need to wait for tens of seconds. It may not sound like a long time, but if more people are using the website at the same time, or the computer performance is poor and cannot run, sometimes it can take a few minutes to generate the result. Not coming out. And after finally producing it, if you find that the picture is different from what you imagined, you have to modify the prompt words again and let the model run again. This goes back and forth until a satisfactory picture is generated, which may take a few hours. And for entry-level users, “how to generate prompt words” is a major pain point in AI drawing. If it takes a long time to generate a picture, patience and interest may have been worn away. Therefore, using the real-time generation function like Krea AI, you can correct the prompt words more immediately and have a higher degree of control over the generated works.

▲ When Midjourney generates a picture, there will be a process from blur to clarity.

▲ LCM is a model developed by the Interdisciplinary Information Institute of Tsinghua University in China. It is officially called the next generation generative model after LDM.

Differences between free and paid functions

(Updated on 2024.03.15) Krea AI currently has 4 plans (as of 2024.02.20), which are free versions, as well as paid Basic, Pro, and Max. The biggest difference between different plans is the usage limit. The generation quota of the free version is updated daily, so people who simply want to try AI drawing or are curious about real-time generated drawings are very suitable to use Krea AI.

How to register for free?

After opening the Krea AI webpage, a “Sign In” sign will appear in the upper right corner. You can choose to directly link your Google account, and then enter the password of the account to log in. There is no need to bind it like the free trial versions of some software. A credit card is required to register. In addition, you can use the real-time generation function and some AI applications on the web without registering an account. However, you will usually be asked to register or log in after a few steps of the web page, and only after logging in can you save the progress of each generation so that you can use it next time. Continue editing. Therefore, it is recommended that you register before using it, and the steps are quite simple and do not take a few minutes.

▲ To log in to use Krea AI, just open the webpage and click the “Sign In” icon that will appear in the upper right corner.

▲ As long as you have a Google account, you can directly select “Continue with Google” to log in. There is no need to register another account.

3 main free features

It is a pity that at the time of writing, the function of Krea AI to generate animated videos is under maintenance and cannot be used. Therefore, the following three major functions that are currently available and free will be introduced and taught, including Realtime Generation to generate images in real time, Upscale & Enhance magnification enhancement tool, and Studio Projects work project.

Real-time Generation uses the LCM model mentioned earlier to generate images in real time. The main function is to quickly generate images. Users only need to use simple prompt words or draw a sketch in the drawing area to generate images. , and the generated results can be adjusted instantly to create a composition that meets your needs. The Upscale & Enhance tool is used to improve the quality of images. It uses AI operations to upgrade the resolution of images. Studio Projects work projects are similar to cloud canvases. Pictures are not generated in real time in this area, but they can be edited online with other people by sharing links.

▲ Krea AI has applications such as instant generation of pictures and improvement of photo quality, but the function of using text to generate animated videos is currently under maintenance.

How to use Real-time Generation to paint?

The first thing I want to introduce is the “Real-time Generation” function of Krea AI that has been mentioned repeatedly before. There are simple text-based drawings and drawing-based drawings, or you can draw a sketch on the drawing board and let AI generate it based on what you draw, or supplement it with various elements and text descriptions. The following uses the “Img2Img” picture to generate a picture of “a young girl wearing white clothes at the beach” to demonstrate several main basic functions. The left is the drawing area, and the right is the system-generated picture. It is important to note that because it is generated in real time, adding or changing any elements may change the entire image.

Step 1: Once you enter the Krea AI homepage, you will see two main functions. Click “Real-time Generation” on the left.

Step 2: There will be default content for the website at the beginning of the “Img2Img” page. Select the “Trash Can” in the lower left corner to clear all elements in the canvas.

Step 3: You can first type the content you want to generate in the text description part. If you have no idea, you can also click the “dice” icon in the lower left corner, and the system will pop up a random prompt word.

Step 4: In the “Colors” category of “Canvas Background”, you can adjust the background color of the canvas.

Step 5: In addition to solid color backgrounds, the system also has many built-in photo backgrounds in the “Photos” category.

Step 6: You can also click “Upload Image” to upload the photo you want, and you can adjust the size of the photo and where to place it on the canvas.

Step 7: Find “Paint Brush” and start painting. You can adjust the color of the brush and change the thickness of the brush through the color pen on the side.

Step 8: In addition to creating with brushes, there are also fixed shapes such as squares, circles, triangles, etc. that can be added. At this time I want to add another sunset, so I choose a circle.

Step 9: Then use the “Select” function key to select the element you want to change. You can adjust its color, size, position, etc. Here I also added “with sunset” to the prompt word to assist AI interpretation.

Step 10: “AI Strength” can adjust the degree of AI intervention. The weaker the intensity, the closer it will be to the appearance of the draft. On the contrary, the stronger the intensity, the more free the AI ​​will be.

Step 11: On the far right, you can also adjust the style of the generated image. In this case, “Concept” is selected. Finally, click “Download Image” and the image will be downloaded to the device.

How to use Upscale & Enhance to upgrade photo quality?

Another major function of Krea AI is to improve the pixels of images. It is not just like many online websites now, where you can simply throw the image in and let the AI ​​judge and generate it on its own. It can also further adjust its style and degree of AI intervention. Wait for adjustments. For example, many pictures captured from the Internet, or photos that have been continuously enlarged and re-enlarged resulting in low quality, can be upgraded through this function. And because the image generated by the previous function also looks blurry, we will directly use this image to demonstrate the Upscale & Enhance tool.

Step 1: Go to the Krea AI homepage as usual, select “Upscale & Enhance” on the right, or click the tab above to enter the “Enhance” page.

Step 2: You can directly drag the image you want to upgrade to the upper left corner of the webpage, or click the “+” in the lower left corner to upload the image.

Step 3: After selecting the photo, the system will automatically identify and generate corresponding prompt words based on the image content. Once confirmed, you can proceed with the upgrade.

Step 4: After upgrading the picture, you can compare the differences before and after the upgrade by sliding the line in the middle of the screen.

Step 5: Or you can change the view mode from the lower right. The left side is the original image before the upgrade, and the right side is after the upgrade.

Step 6: The image resolution can be enlarged from 2 times to 8 times, and can be processed up to 2048 pixels.

Step 7: There are various styles to choose from when upgrading photo quality. However, after testing, there is not much difference in the upgraded pictures when choosing different styles.

Step 8: You can also use “AI Strength” to adjust the degree of AI intervention. The degree of intervention here is more like the difference between blurry and clear pictures.

Step 9: After the actual measurement, you can use positive and negative prompt words, AI intervention degree and other numerical values ​​to adjust the elements of the picture, such as the color of clothes. However, the effect is not very good, so it is not recommended.

Step 10: It doesn’t matter if you prefer the previous version after generating it. Each time you generate the result, it will automatically be saved in the list below. You can also see the various values ​​​​set originally by clicking on it.

Step 11: Finally, click on the image you want to save below, press the “Download” button in the upper right corner of the web page, and the image will be downloaded to the device.

How do I use Studio Projects and share my canvas?

It is worth mentioning that when generating images on Studio Projects, there are many different AI models to choose from, and there is no need to download the model to the device. You can use it directly by opening the web page. Not only is there no hardware limit, even if it is a document It can also be used on a computer and does not take up hard drive space. The following will introduce how to create a canvas, generate pictures and other basic functions in Studio Projects, and share the link with others to edit together.

Step 1: Go to the Krea AI homepage, scroll down, and click “New canvas Project” in the Studio Project column to add a new canvas.

Step 2: Then you can use the prompt words to create a diagram. If you have no idea, you can also click the “dice” icon and let the system generate it randomly.

Step 3: In addition, you can select a specific AI engine in the canvas to generate pictures of different styles. There are built-in AI models and models trained by others in the community.

Step 4: In the advanced settings, you can adjust the image size, diffusion steps, negative prompt words, etc.

Step 5: Then press “Generate” to generate two versions of the image below. Click on the image to add it to the canvas on the right for further editing.

Step 6: If you feel that the picture is incomplete and you want to extend it in any direction, up, down, left, or right, click on the picture first, and then select “Expand” on the upper toolbar.

Step 7: After entering the prompt word, you can extend the photo. The system will also generate two pictures. You can add them to the canvas with just one click. The prompt words can maintain the content of the original picture, or add more descriptions.

Step 8: If you want to change the color tone of the entire image, select the “Recolor tool” on the toolbar. You can click on the image to call up the toolbar above, or you can find it in the column on the left.

Step 9: Click the “Change Button” and the system will automatically pop up different color combinations. After selecting a satisfactory color match, press “Recolor”.

Step 10: In addition, if you want to know the information about the picture, click on the picture and the “Information” icon will pop up in the upper right corner of the web page. Click it to view it.

Step 11: At the top, you can edit the project name of the canvas. If you want to share the canvas for others to edit together, click “Share” and the link will be automatically copied and you can share it.