Filmvert is setup to operate on a roll-by-roll basis. This presents a few advantages, namely that the exact same inversion settings can be used across all images in a roll.

Several images within a roll can be analyzed, and the best/most accurate analysis can be copied across some/all images in a roll.

Start by either importing individual images, or a directory containing images. Individual images will need to be added to an internal 'roll' with a given name and working directory. This working directory is where inversion settings and metadata will be saved.

Directories imported will create rolls based on the directory name, and all images imported.

The "Image Colorspace Settings" are relevant to non-camera raw files. It is preset to the most common default (sRGB), but other options are available should they be needed. More on the color pipeline can be found in the Technical Documentation.

The first step in analyzing an image is sampling the base color. This is achieved by holding Cmd/Ctrl + Shift, and clicking and dragging an area in the image viewer to select a region of the image that is purely film base. If the borders of the film aren't available (in the case of Pakon scans) a nearly/pure black area in the negative could be sampled, or the base color left at the default setting.

This function serves purely to bias out the base color so the analysis produces values closer to black/white, rather than biased against the base color. It is entirely possible to get the same result by both sampling, and not sampling the base color.

By default, the Analysis Region is set to 90% of the image size. This should be adjusted to encompass the image, as well as some film base (if available). The analysis will only pay attention to pixels within the bounds of the region. This means that this can be used to influence the analysis by excluding/including specific portions of the image.

In the analysis process, the image is blurred, and then the lightest and darkest pixels are found by calculating the luma value of each pixel. The purpose of the blur is to smooth out any extraneous pixel values that may throw the analysis off. The Analysis Bias is used to control the blur, and different results can be had based on the value set here.

The RGB values for the minimum, and maximum luminence pixels are used as the Analyzed Black Point and Analyzed White Point.

The individual RGB channels of the Blackpoint/Whitepoint can be adjusted to tweak the inversion of the image. If the film base is included in the analysis region, the blackpoint is usually located somewhere in the base and is for the most part a good sane value. The whitepoint usually needs some slight adjustments.

The individual RGB channels can be adjusted, or in the color picker popup, the Hue and Saturation can be adjusted to give a nicer balance between cyan/magenta/yellow/green.

The Min/Max points in the image viewer can also be dragged to adjust the analysis:

The Grade section allows for some further tweaks to be made to the inversion to further refine it

Image settings can be copied (Cmd/Ctrl+C), and pasted across one or more images to view the analysis across more of the roll. Other images can be analyzed as well, and their settings copied and pasted across further images.

Settings can be synced across an entire roll using the menu option Roll->Sync Roll

It's often the case that a few images need to be analyzed to find a single global inversion that works across all images.

Do keep in mind that the inversion process is objective. Meaning with the same settings, the exact same operation is happening image to image. For rolls shot in a single lighting setup/location, this means a single inversion setting may work across all images. But for a roll containing different lighting conditions, various inversion settings may be needed in order to get a good subjective inversion out of each image. There is no right answer other than what you think looks the best.

There are a handful of custom metadata fields available for taking notes while inverting, including camera details, locations, film stocks, dev&scan info. This metadata is packaged with the save-data from the inversion and included in the sidecar, and final output images.

This metadata can be exported as a CSV for external metadata workflows.

Image settings (and metadata) can be saved out. This data is saved into an xmp sidecar file alongside the source image. Full-roll metadata can be saved out (Save Roll). This saves as a fvi file in the roll directory. This fvi file is able to be re-imported and applied to a roll.

Image settings from the xmp file are automatically loaded when re-importing images/rolls.

Unsaved changes will appear as a purple dot next to the image name in the thumbnail viewer.

Once there are image(s) that are considered done, they can be exported with File->Export Image(s) or File->Export Roll(s). If exporting rolls, the option to select which of the open rolls to export will appear in the popup.

There are several basic file formats available for export, as well as a couple of options for each file format.

The same colorspace options are available in the export window as in the import window. This allows the images to be saved out in other colorspaces for further processing.

The option to bake in any crops and rotations is enabled by default. There is also the ability to bake in a border on exported images.

Files are exported with the same name as the source file. All metadata fields, as well as the inverstion settings are packed into the "Image Description" of the resulting file.