Referencing Images And Other Files In Power BI Reports Using OneLake URLs

Several years ago I wrote a very popular blog post about how to store images for your reports inside your Power BI semantic model. It solved the problem of how you could use display images (for example of products) inside your reports without making those images available via a public URL or personal OneDrive Embed Codes. I was very proud of how efficient the M code to do this was but the code was complicated and storing images as text inside a semantic model makes refreshes a lot slower and increases the size of your semantic model in memory, so it’s not ideal. The good news is that, if you have enabled Fabric in your tenant, the August 2026 release of Power BI brings a much better way of solving this problem: you can now store your images (and indeed other files) inside OneLake and reference them from there. This means you can store your images in a secure location, alongside all of your other data, and make them available for use in Power BI. What’s more this doesn’t just work for images, it also works for other types of files such as GeoJSON files used by map visuals.

To illustrate how this works, I created a semantic model containing data from the 2024 UK general election. The details of the model don’t matter that much except for the fact that there was a column called Country Name containing the distinct values England, Scotland, Wales and Northern Ireland, and a measure called Total Electorate. I published the model to a workspace and then, in the same workspace, created a lakehouse. In the Files section of the lakehouse I created a new folder and then uploaded five jpg files of the flags of England, Scotland, Wales and Northern Ireland as well as the Union Jack.

I made a note of IDs of the workspace and lakehouse. Then, in the semantic model, I created a measure with the following definition which returned the url of the image in my lakehouse for the selected country or, if no country was selected, returned the url of the Union Jack:

Flag =
IF (
    HASONEVALUE ( 'Election'[Country name] ),
    "https://onelake.dfs.fabric.microsoft.com/insertworkspaceidhere/insertlakehouseidhere/Files/Flags/"
        & SELECTEDVALUE ( 'Election'[Country name] ) & ".jpg",
    "https://onelake.dfs.fabric.microsoft.com/insertworkspaceidhere/insertlakehouseidhere/Files/Flags/United Kingdom.jpg"
)

I then set the Data Category property of this measure to be Image URL:

Finally I create a report with a table visual in, dragged in the Country Name column, the Flag measure and the Total Electorate measure, and saw the flag images displayed in the table:

I then uploaded a GeoJSON file containing the outlines of the countries of the United Kingdom from the Office for National Statistics Open Geography Portal to another folder called GeoJSON in my lakehouse:

I viewed the Properties for the file and copied its URL:

I then created a Shape Map visual, dragged the Country Name and Electorate measures into it, went to the Map Settings properties and set the Type to “URL” and pasted in the URL of the GeoJSON file from OneLake into the Enter A URL box. This gave me a Shape Map showing the outline of the UK and colouring representing the size of the electorate:

[Note: at the time of writing this doesn’t work with the Azure Maps visual but this will be fixed soon]

Nothing here in terms of displaying images or using GeoJSON files with maps in Power BI reports is new. What is new, as I said, is where the images and the GeoJSON files are stored. The ability to store these files in OneLake means no more workarounds are needed to work with dynamic images as in the flag example, and it also means that you never need to upload a GeoJSON file containing sensitive data used by a map visual – you just need to store the GeoJSON file in OneLake and if you need to change it you simply replace the file without having to edit the report. And of course you can now automate the loading of these files into OneLake in the same way that you automate the loading of other types of data. If you’re a Power BI shop that hasn’t enabled Fabric yet then this is one more reason to do so.