File Size Doubled After Compressing and Reprojecting Drone Image in ArcGIS Pro

[logri-sl]

Problem Summary

I'm trying to compress a drone raster image while converting it from EPSG:32628 (WGS 84 / UTM Zone 28N) to EPSG:4326 (WGS 84 Geographic) in ArcGIS Pro, but the output file size is double the original file size instead of being smaller.

My Workflow

I used the Copy Raster tool with the following settings:

1. Input Raster: Freetown drone image (EPSG:32628)
2. Output Coordinate System: EPSG:4326 (WGS 84)
3. Output Format: TIFF (.tif)
4. Pixel Type: 8-bit unsigned
5. NoData Value: 0
6. Compression Type: JPEG
7. Compression Quality: 75
8. Build Pyramids: Checked
9. Pyramid Compression: JPEG
10. Resampling Technique: Bilinear

Expected Result

A compressed image with significantly reduced file size (50-80% reduction) in EPSG:4326.

Actual Result

The output file size is approximately 2x the original file size.

Questions

1. What might be causing the file size to increase instead of decrease?
2. Could the CRS conversion from UTM (meters) to Geographic (degrees) be increasing pixel count?
3. Are internal pyramids adding significantly to the file size?
4. Is there a better workflow to compress AND reproject simultaneously without increasing file size?
5. Should I separate the reprojection and compression into two separate steps?

Environment

• Software: ArcGIS Pro
• Input Image: Drone imagery (RGB)
• Original CRS: EPSG:32628
• Target CRS: EPSG:4326

What I've Tried

• Using Copy Raster tool with JPEG compression
• Setting compression quality to 75
• Building pyramids with JPEG compression

Any advice on the proper workflow or settings to achieve both compression and CRS conversion without increasing file size would be greatly appreciated!

[tikola]

I see at least two possibilites:

1. When square file is reprojected image gets tilted and truly amount of rows and columns increase. Check input and output row and column amount and do you get black triangels in image corners. So is your image size (row&column amount) truly growing after reprojection?
2. Another option - input TIF had some compression that was better than jpeg you are now using. TIF can contain so many different things that there well might be some jp2 or some other stronger compression than the one you now use. However that would most likely be something way bigger that 50-80 %

[JudithWammack]

@logri-sl  have you figure out a solution on this? I am running into the same issues with file size doubling. 

[logri-sl]

No, unfortunately, I wasn't able to resolve this issue. Good luck! I hope you're able to do so!

[JudithWammack]

@logri-sl Have you worked with the Cloud Raster Format (CRF) before? We recently came across this format and found that it significantly reduces file size while maintaining imager quality.

Our current workflow involves taking 536 aerial TIFF images, converting them into a more compressed format, reprojecting them, and then publishing them to create a cache. So far, using CRF has produced smaller file sizes without compromising image quality.

Do you think this approach would work for your needs?

For reference, the following documentation provides an overview of CRF workflows and their capabilities:
https://doc.arcgis.com/en/imagery/workflows/standard-workflow/preparing-your-data/data-sources-and-f...