Skip to content

Questions

The awkward ones first.

Including what single-image mode genuinely cannot do, why there is no DNG export, and what the free plan is for.

How the extraction works

What does PresetForge actually do?
It reads a colour transform out of your images and writes it as a Lightroom preset. Given the original and the edited version of the same photo, it compares them pixel by pixel and derives the per-channel tone curves, the per-band hue, saturation and luminance shifts, the split colour grading, the grain and the vignette. Given a single image, it analyses that image's tonal and colour statistics and produces a preset that pushes a neutral photo toward them.
What is the difference between before/after mode and single-image mode?
Before/after mode measures. Because both states of the same frame are available, the transform between them is a fact that can be read directly off the pixels, and the result reproduces the edit closely enough to report a residual error. Single-image mode infers. With only one image there is no way to separate the edit from the scene — a warm photo might be a warm grade or a sunset — so the result is a well-founded starting point rather than a measurement.
Is before/after mode a paid feature?
No. Exact before/after extraction is available on the free plan, alongside .xmp export and eight-anchor tone curves. Gating the feature that makes this worth using would mean nobody could find out whether it works. Pro adds the full colour panel — Colour Mix, Colour Grading and grain — sixteen-anchor curves, .cube and .json export, unlimited extractions and batch apply.
Is there any AI involved?
No. The extraction is arithmetic: binning, medians, monotone curve fitting and colour-space conversion. That is why results are identical every time you run the same pair, why it takes seconds, and why there is no per-request cost that has to be passed on to you.
How accurate is it?
In before/after mode, accuracy is reported rather than claimed. After deriving the curves, the engine applies them back to the original and compares the result to the target, reporting the mean absolute residual per channel on a 0–255 scale. A residual under about 3 means a reproduction that is very hard to tell from the target. Looks that depend heavily on local adjustments will score worse, because a global preset cannot express them.
Do the two images need to be the same size?
They need to be the same frame and the same crop. Different pixel dimensions are fine — the images are resampled to a common analysis size — but if one is cropped and the other is not, the pixel correspondence breaks and the result will be meaningless.

Formats and compatibility

What formats can I export?
Three. A .xmp preset for Lightroom Classic, Lightroom desktop, Lightroom mobile, Adobe Camera Raw and Bridge. A 33×33×33 .cube 3D LUT for Premiere, DaVinci Resolve, Final Cut Pro, OBS and most camera monitors. And .json, the raw measured numbers including every curve anchor, for inspection and scripting. .xmp is on every plan; .cube and .json are Pro.
Why is there no DNG export?
Because a DNG preset is not a preset — it is a real DNG image carrying Develop settings in its metadata, and installing it means importing a junk photo and copying its settings off. It existed only because Lightroom on iOS could not import .xmp files. It has been able to since 2024, on both iOS and Android, so a DNG would be a worse file requiring more steps.
Will the .xmp import into Lightroom without a conversion step?
Yes. The files are written the way Lightroom Classic writes its own: a bare x:xmpmeta document with no byte-order mark and no xpacket wrapper, the correct crs:Supports flags, colour grading written with the legacy SplitToning attribute names Adobe actually uses, and bipolar sliders written with explicit signs. It appears as an ordinary preset.
How is white balance written?
As the relative Incremental pair rather than an absolute temperature. That is the form that applies correctly to already-rendered files such as JPEGs, and it means the preset offsets whatever white balance a photo already has instead of overriding it — which is what you want if the preset is going to be used on more than one photograph.
Does it work with Capture One, darktable or Affinity?
Not through .xmp — that format is specific to the Adobe raw pipeline. The .cube LUT will load anywhere that reads 3D LUTs, which includes Capture One and darktable, though you will lose the spatial parts of the look such as grain and vignette.
What image formats can I upload?
JPEG, PNG, WebP, TIFF and AVIF, up to 25 MB per file.

Plans and billing

What is included in the free plan?
Five extractions a month, both single-image and exact before/after modes, .xmp export and eight-anchor tone curves. No card required.
What does Pro add?
Unlimited extractions, the full colour panel — Colour Mix, Colour Grading and grain — sixteen-anchor tone curves, .cube LUT and .json export, and batch apply across up to 50 photos.
What is the Lifetime plan?
Everything in Pro, paid once. It is the same feature set; the difference is billing.
Why are eight-anchor curves the free limit?
Eight anchors describe almost any real edit accurately. Sixteen describe it more tightly, which is better when the source pair is clean and worse when it is noisy — a curve fitted too closely to one image's quirks will posterise on the next one. The cap is a genuine ceiling on precision, not an artificial restriction on something free users would otherwise get.
Do unused extractions roll over?
No. The free counter resets monthly.

Using the presets you make

How do I install the .xmp in Lightroom?
In Lightroom Classic, Develop module, click the + on the Presets panel and choose Import Presets. In Lightroom desktop, File → Import Profiles & Presets. In Lightroom mobile, open a photo, tap Presets, tap the three-dot menu and choose Import Presets. Full walkthroughs are in the guides.
Why does my preset look different on a different photo?
Because a preset stores instructions, not an appearance. Different starting exposure, different white balance, raw versus JPEG, a different camera profile, different subject colours or clipped channels will all change what the same instructions produce. The guide on this covers all six causes and what to do about each.
Can I edit the preset afterwards?
Yes, and you should. A measured preset is a set of ordinary Develop settings — open it in Lightroom and change anything. The .json export gives you the same numbers in a form you can script against.
Can I sell presets I make with PresetForge?
Presets you derive from your own photographs are your own work and yours to sell. Presets derived from someone else's images raise the same questions any derivative work does, and naming a preset after a film, brand or photographer is a trademark question independent of the colour science.

Privacy

What happens to the photos I upload?
They are processed to produce your preset and then deleted. They are not used to train anything — there is nothing to train, because the engine is arithmetic rather than a model.
Do you keep my presets?
Your extraction history is stored against your account so you can download a preset again later. You can delete any of it at any time.

Still unsure? It is free to find out.

Five presets a month, exact before/after mode included. Upload a pair of your own photographs and read the curve it gives you.