[REQUEST] Linear scaling of raw2img conversion

[TheSnowCat]

Type of Request

• Functionality
• UX/UI
• Documentation
• Quality of Life
• Other

Is your feature request related to a problem? How pervasive is the problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like
option to do raw2img with linear scaling of png values from min to max -> 0-256

Describe alternatives you've considered
imageJ

Additional context
need to figure to whether the scaling is min to max of entire raw or slice by slice (entire raw makes the most sense)

[tparkerd]

@TheSnowCat Is this still needed? Out of curiosity, why do you need to scale the data?

Scaling should be relatively straightforward, so would there be any other types of scaling that you would need (e.g., log or clipping)?

[kenanoestreich]

Long-winded answer incoming.

Yeah we're trying to figure exactly what we want to do. The problem is that TopoRoot uses Fiji as a library to load in .raw files and convert them to image stacks. But in doing so, Fiji has some auto-contrast-enhancing functionality that scales the histogram to a subsection (that ideally contains "the most relevant data"). So when we try to run TopoRoot on our image stacks that haven't been scaled, we (Owen and Alex) have been running into problems with TopoRoot needing the 3 thresholds to be a bit farther apart from each other. That's kinda the jist of it. I've tried a bit to see how ImageJ determines its "display range" that it uses to adjust the histogram so we could replicate it, but I haven't been able to find it in the docs or source yet. I'm pretty against adding Fiji as a library in rawtools; that's what TopoRoot does.

Also Hi buddy haha

[tparkerd]

Hey! 🧟

I didn't know it did that. Just so I'm on the same page, are you talking about the downsample macro that runs to preprocess the .raw (i.e., macros/downsampled.ijm or macros/downsampleRaw.ijm)? If not, where is this happening?

Have you directly compared the slices from raw2img and ImageJ's output slices? If so, how do they differ? How do they differ without resizing the images? In other words, just splitting the .raw into slices—no interpolation involved.