
Cocreate began like many projects do: with a random idea that came up over drinks with friends. We thought it would be cool to build a project where people could combine two of their NFTs and produce a new hybrid output. For example Cool Cat #X plus SupDuck #Y equals Cocreation #Z.
The idea quickly turned into research, which turned into a prototype, which turned into a development sprint, which turned into live deployment. The solution we came up with was to train a machine learning model for “generative style transfer” where the algorithm takes in two pictures and creates a third one based on the content of the first and the style of the second. This approach is based on a series of academic papers and open source tech that other nerds can read about in the footnotes below. [1]
In trying to make this as high-quality a model as we could we trained the algorithm with everything from ancient art, renaissance art, and modern art, to digital art and several NFT collections. At this point, the outputs are fun and often surprising. See an example of a Bored Ape base remixed with a SupDuck style below.

Today, we’re excited to share that we’re opening up the cocreate model for everyone to use! Why is this interesting? Usually, in the Ethereum ecosystem a user mints some amount of NFTs from a new project, then they receive random outputs they didn’t have any control over making. By contrast, Cocreate invites users to be… co-creators.
The idea is everyone can mint a “canvas,” then with a second transaction a new piece of art is applied to the canvas (now a “creation”) based on the given inputs. Potential outputs will be viewable before the second transaction to make sure people are happy with their creation in advance of paying gas.
There are 6969 canvases available for mint at 0.069420 ETH each, for the culture. Canvas holders are able to connect their wallet to the cocreate website, test possible NFT combinations, then mint any creations they find appealing. The only fee associated with the second step is to pay Ethereum’s gas costs.
We decided to split the canvas NFTs and the cocreation process itself to create an anti-FOMO buffer period. If this whole process took place in one go, it’s possible people would rush to just get a canvas for speculative reasons as opposed to being thoughtful about what NFTs they would like to use to create something unique. Having a step between the minting and the cocreating allows people to engage with our team and the community via Discord, just to make sure everyone understands what is happening.
For those who are interested, the contract can be found here and the site can be found here. On-chain sleuths will notice two things. First, we have kept 50 canvasses for community giveaways, the first of which people will hear about soon! Second, the contract is actually called MashNFT, which was the original name we had for the beta version of the project when we launched back in September.
We look forward to seeing what people produce from our algorithm (and *pssst* other custom algorithms in the future).
If you have any questions or would like to chat further, join the Discord and we would be happy to connect.
-Cocreate team
[1] https://arxiv.org/abs/1508.06576, https://arxiv.org/abs/1705.06830

Cocreate began like many projects do: with a random idea that came up over drinks with friends. We thought it would be cool to build a project where people could combine two of their NFTs and produce a new hybrid output. For example Cool Cat #X plus SupDuck #Y equals Cocreation #Z.
The idea quickly turned into research, which turned into a prototype, which turned into a development sprint, which turned into live deployment. The solution we came up with was to train a machine learning model for “generative style transfer” where the algorithm takes in two pictures and creates a third one based on the content of the first and the style of the second. This approach is based on a series of academic papers and open source tech that other nerds can read about in the footnotes below. [1]
In trying to make this as high-quality a model as we could we trained the algorithm with everything from ancient art, renaissance art, and modern art, to digital art and several NFT collections. At this point, the outputs are fun and often surprising. See an example of a Bored Ape base remixed with a SupDuck style below.

Today, we’re excited to share that we’re opening up the cocreate model for everyone to use! Why is this interesting? Usually, in the Ethereum ecosystem a user mints some amount of NFTs from a new project, then they receive random outputs they didn’t have any control over making. By contrast, Cocreate invites users to be… co-creators.
The idea is everyone can mint a “canvas,” then with a second transaction a new piece of art is applied to the canvas (now a “creation”) based on the given inputs. Potential outputs will be viewable before the second transaction to make sure people are happy with their creation in advance of paying gas.
There are 6969 canvases available for mint at 0.069420 ETH each, for the culture. Canvas holders are able to connect their wallet to the cocreate website, test possible NFT combinations, then mint any creations they find appealing. The only fee associated with the second step is to pay Ethereum’s gas costs.
We decided to split the canvas NFTs and the cocreation process itself to create an anti-FOMO buffer period. If this whole process took place in one go, it’s possible people would rush to just get a canvas for speculative reasons as opposed to being thoughtful about what NFTs they would like to use to create something unique. Having a step between the minting and the cocreating allows people to engage with our team and the community via Discord, just to make sure everyone understands what is happening.
For those who are interested, the contract can be found here and the site can be found here. On-chain sleuths will notice two things. First, we have kept 50 canvasses for community giveaways, the first of which people will hear about soon! Second, the contract is actually called MashNFT, which was the original name we had for the beta version of the project when we launched back in September.
We look forward to seeing what people produce from our algorithm (and *pssst* other custom algorithms in the future).
If you have any questions or would like to chat further, join the Discord and we would be happy to connect.
-Cocreate team
[1] https://arxiv.org/abs/1508.06576, https://arxiv.org/abs/1705.06830
Cocreate is a generative art platform where users can collaborate with AI to make new NFTs based on ones they already own. Our machine learn
Cocreate is a generative art platform where users can collaborate with AI to make new NFTs based on ones they already own. Our machine learn
Share Dialog
Share Dialog

Subscribe to Cocreate

Subscribe to Cocreate
<100 subscribers
<100 subscribers
No activity yet