Freepik Spaces was announced right before DevLearn, and I finally had a chance to spend a few hours testing it out. Spaces is a node-based tool for generating images and video. In a node-based tool, you can combine multiple tools into a single workflow and map out multiple versions. This can be especially helpful for media generation where it’s common to switch back and forth between multiple tools. I can definitely see the potential for improving efficiency and consistency with a node-based tool like Freepik Spaces. This could also be a way to save money because you only pay for one tool, but you get access to a bunch of different options within it.
As with several of my previous posts, this is more about my first impressions after experimenting with a tool rather than a complete how-to or tutorial. My tests mostly focused on generating character images, since that’s one of my most common use cases for AI image generation.
Freepik Spaces templates
Freepik has a growing collection of templates for Spaces. These give you some example workflows to start from and adapt. As you can see in the image below, there are templates to “Get alternate angles” and “Modify facial expressions.” Many of the templates are more aimed at marketing, but some of these might work for learning too.
I decided to start with the “Character generator” template. You can see how different this is for working than typing prompts in a chat window. You do still use some text prompts here, but you also use reference images or other media.
This template includes several steps, which are repeated for each of the four characters included.
Describe your character (the Input).The AI assistant generates an image prompt based on your description.The Output is 4 generated images.You pick the best image and connect it to the next part of the process.You describe a different background that your character will be placed in.The AI assistant writes a prompt to put the character in that background.It generates an image of the character in the new background.It generates a video of the character.
The characters in the template all look like models, but I wanted something more realistic. I changed the prompt to customize my character.
Create a character with this description:
An Alaskan Native womanAge: Early 30sFace: round face, brown eyesHair: Black, one long braidClothes: patterned red coat over a black T-shirtStyle: casualExpression: slight smile, listening
This prompt is then connected to image generation nodes. This template uses Nano Banana Pro for image generation and Kling for video generat, but you could pick different models. This means it’s easy to use one tool to generate the initial images and then switch to something else for edits.
You can see the output for the second image of the generations and the changed background. Nano Banana Pro does a great job with these types of edits.
While I don’t usually need a video like this for training, I could see this one being part of an intro or B-roll footage.
Starting from a blank template
I also tried starting from a black template and creating my own space. I used this video on Freepik Spaces to help me get started. This time, I wanted to focus on generating different expressions and emotions for a character and then changing the angle.
This time, I switched from using NanoBanana Pro to using Flux Kontext Pro for the image generation to save on credits. I think Flux did OK, although the skin is a little shiny even after upscaling. It’s good enough for an experiment, but if this were for a real client project, I probably would have refined the prompt for the woman a little to improve the lighting and skin texture. Flux did a good job of keeping the facial features consistent while changing the expressions, so I was happy with those results.
I also used Magnific to upscale the image once I picked my favorite one out of the set. Being able to switch from the image generation tool to the upscaler seamlessly was a definite plus here.
I used the same prompt for each of the expressions, but I edited the emotion and description each time. I did use the AI tool to suggest improvements to the prompt, but I ended up just copying that text manually into multiple different prompts. Is there a more efficient way to create a series of similar prompts like this? Maybe, but I didn’t find it in my quick testing. I grouped all of the expressions together so it’s easy to find them.
The best thing about this is that once I created this flow for one character, I could just copy and paste the structure for another character. I adjusted my initial character description and then ran all of the subsequent nodes again. For a few images, I didn’t like the results, so I manually edited and ran those images again. The repeatability of the process is one of the big potential advantages of a node-based tool like this.
Cost and Plans for Freepik Spaces
Spaces isn’t available on Freepik’s free plan, so I decided to pay for a single month of the Essential plan. That’s currently $9/month. They have several higher plans with discounts for annual subscriptions. The Essential plan was OK for testing it out, but I burned through most of my credits in about 3 hours. If I was really using this tool for client projects, I’d probably need to upgrade to a Premium ($20/month or $144/year) or Premium+ ($39/month or $294/year) plan.
Still, $20 for a month for pilot or single project is reasonably affordable. If it helps you consolidate several other subscriptions for other tools, it’s easier to justify the cost. I certainly paid over $300/year for stock images in the past, and that’s part of my calculation for cost with image tools.
But right now, I’m also paying $200/year for Poe, which also includes most of the same image and video generation tools. The workflow in Freepik is smoother since it’s more visual and requires less switching. I can see how it could save me time, which is worth paying for, but I’ll probably wait to have a client project that really needs it. I also want to spend more time testing Flora and try out Weavy, two other similar node-based tools.
Final thoughts
Working in a node-based tool like this felt sort of slow initially, but I think that’s more about my unfamiliarity than an actual problem with the tool. I also think that it takes time to set up the initial workflow, but I’d save time repeating the process and reusing prompts and processes. Having multiple tools in one place and being able to easily switch between tools are big benefits too.
I’m not ready to commit to an annual plan on Freepik yet because I want to test the other node-based tools first. However, this definitely seems like it has potential for generating consistent character images.
Compare with other tools
Compare how Freepik Spaces works with other tools for generating character images.
I also wrote a longer post comparing the strengths and weaknesses of various AI image generation tools.
