what is bit depth and why does it matter for editing

You have pushed a sky gradient three stops in Lightroom and suddenly smooth blue turns into ugly stair steps. That is what is bit depth and why does it matter for editing in a single frustrating moment. Bit depth is the number of tonal steps available per color channel.

More steps mean smoother gradients and more editing headroom before the image falls apart.

The typical JPEG gives you 256 levels per channel while a 14‑bit RAW file delivers 16,384 levels per channel. Per ITU‑R BT.2100 the minimum for HDR mastering is 10‑bit which provides 1,024 levels per channel. Let us walk through the three bit depth decisions that control whether your edits hold together or break.

Quick Answer

Bit depth is the number of tonal values per color channel. Eight bit gives 256 values. Fourteen bit gives 16,384 values.

Higher bit depth prevents banding when you push exposure or grade shadows. You need high bit depth at capture and during editing. Delivery can be lower with proper dithering.

Why Your Gradients Look Like Stair Steps (And What Bit Depth Has to Do With It)

You have seen it. A clean blue sky turns into visible bands after a curves adjustment. A dark shadow lift reveals concentric rings instead of smooth tone.

That is quantization error. The file simply ran out of tonal steps to represent the gradual change. Eight bit color spaces only have 256 steps per channel.

Stretch those steps and gaps appear. Fourteen bit RAW starts with 16,384 steps. The gaps are microscopic.

You can push farther before the stairs show up. Video log curves like S‑Log3 or C‑Log compress dynamic range into fewer code values. Ten bit recording preserves 1,024 steps.

Eight bit log discards most of them before you even start grading.

What Bit Depth Actually Means — Without the Math Degree

Bit depth defines how many discrete values each color channel can hold. One bit is two values (black or white). Eight bit is 256 values.

Ten bit is 1,024 values. Fourteen bit is 16,384 values. Sixteen bit is 65,536 values.

Thirty two bit float uses scientific notation so the range is effectively infinite. Each step represents a tiny change in luminance or chroma. More steps equal finer granularity.

Think of a ruler. Eight bit has marks every inch. Fourteen bit has marks every hair width.

When you stretch the ruler the coarse marks separate into visible gaps. The fine marks stay smooth. That is the entire concept.

No calculus required.

The Three Bit Depth Decisions You're Making (Whether You Know It or Not)

Every imaging pipeline involves three distinct bit depth choices. First is capture bit depth. This is what the sensor records.

Second is working bit depth. This is the precision your software uses for calculations during editing. Third is delivery bit depth.

This is what the final file contains. They do not have to match. A smart pipeline captures at 14 bit.

Works in 16 bit or 32 bit float. Delivers at 10 bit for HDR or 8 bit with dithering for SDR web. Mistakes happen when you confuse them.

Shooting 14 bit RAW then editing in an 8 bit timeline throws away the capture advantage. Editing in 16 bit then exporting 8 bit without dithering bakes in banding. Each decision point has different requirements.

Capture Bit Depth: What Your Camera Gives You

Consumer cameras typically record 12 bit RAW for stills. Pro bodies push 14 bit. Medium format reaches 16 bit.

Video is different. Most mirrorless cameras record 8 bit 4:2:0 internally. Higher models offer 10 bit 4:2:2.

Cinema cameras like ARRI Alexa shoot 12 bit or 16 bit RAW. Log curves change the math. A log curve allocates code values to preserve highlights and shadows.

Ten bit log gives you 1,024 steps across 13+ stops. Eight bit log gives you 256 steps across the same range. That is why 10 bit internal recording matters for video.

Check your camera menu. Set the highest bit depth option. For stills choose lossless compressed RAW if available.

For video enable 10 bit 4:2:2 if your card write speed supports it. As of 2026 most mid range bodies from Sony Canon Nikon Fuji offer 10 bit internal.

Working Bit Depth: Where Your Edits Live

This is the precision your editing application uses for every calculation. Photoshop defaults to 8 bit for legacy compatibility. You must change it to 16 bit or 32 bit per channel.

Lightroom and Capture One process RAW in a 16 bit linear pipeline internally. DaVinci Resolve defaults to 32 bit float timeline. Premiere Pro defaults to 32 bit float if you enable it.

After Effects defaults to 32 bit float. The rule is simple. Work at equal or higher bit depth than your capture.

Never downconvert before you finish editing. Every curves move every layer blend every LUT application accumulates rounding errors in 8 bit. In 16 bit those errors stay invisible.

In 32 bit float they effectively do not exist. High bit depth working space also prevents clipping during extreme adjustments. You can push exposure plus four stops in 32 bit float.

Pull it back. The data is still there. Try that in 8 bit and you get posterized noise.

Delivery Bit Depth: What Your Audience Sees

Delivery bit depth is the final container. Web standard is still 8 bit sRGB JPEG or HEIF. HDR delivery requires 10 bit minimum per SMPTE ST 2084.

Dolby Vision mastering demands 12 bit. Print labs accept 8 bit sRGB JPEG but fine art shops prefer 16 bit Adobe RGB TIFF. The key rule: never downconvert before the final export.

Keep the master at working bit depth. Export a copy at delivery spec. Apply dithering on the 8 bit export.

Dithering adds controlled noise that masks quantization bands. Most exporters handle this automatically if you enable it. For 10 bit HDR delivery use HEVC or AV1 in a Rec.2020 container.

Do not dither 10 bit to 10 bit. Do not dither 16 bit to 16 bit. Only dither when reducing bit depth.

Decision Branch: Stills Photographer Workflow

If you shoot stills start with camera RAW at maximum bit depth. Set Lightroom or Capture One to 16 bit ProPhoto RGB working space. Edit non destructively.

Every slider move stays parametric. Export 16 bit TIFF for archive. Export 8 bit sRGB JPEG with dithering for web.

Export 16 bit Adobe RGB TIFF for print. If you composite in Photoshop keep the document in 16 bit or 32 bit. Only flatten at final output.

Never edit a JPEG then save as JPEG again. That doubles the quantization damage. If a client demands 8 bit deliverables give them the dithered export.

Keep the 16 bit master for future revisions.

Decision Branch: Video Editor / Colorist Workflow

If you grade video the timeline bit depth decides your latitude. DaVinci Resolve defaults to 32 bit float timeline. Keep it there.

Premiere Pro needs 32 bit float enabled in sequence settings. Final Cut uses 32 bit float internally. Import 10 bit or 12 bit log footage.

Apply color space transform to working space like DaVinci Wide Gamut or ACEScg. Grade in 32 bit float. Apply output transform for delivery.

Export 10 bit ProRes 422 HQ or 12 bit ProRes 4444 for master. Export 10 bit HEVC for HDR streaming. Export 8 bit H.264 with dithering for SDR web.

Never grade in an 8 bit timeline. Log footage crushed into 8 bit loses shadow separation permanently.

Decision Branch: Hybrid Shooter (Stills + Video) Workflow

Hybrid shooters face two pipelines. Stills want 14 bit RAW into 16 bit ProPhoto. Video wants 10 bit log into 32 bit float timeline.

The camera settings conflict. Most mirrorless bodies let you set stills to RAW and video to 10 bit 4:2:2 separately. Do that.

Do not shoot video in 8 bit to save card space. The grading penalty is not worth it. Use separate ingest folders.

Process stills in Lightroom. Process video in Resolve. Keep separate masters.

Deliver stills as 8 bit JPEG or 16 bit TIFF. Deliver video as 10 bit HDR or 8 bit SDR. If you pull frames from video for stills treat them as 10 bit source.

Upsample to 16 bit working space in Photoshop. The frame grab will never match true RAW still quality.

Decision Branch: Print vs Screen vs HDR Delivery

Print demands 16 bit Adobe RGB or ProPhoto TIFF. Soft proof with the lab ICC profile. Screen SDR delivery is 8 bit sRGB JPEG or HEIF with dithering.

HDR delivery is 10 bit HEVC in Rec.2020 PQ or HLG. Dolby Vision is 12 bit with dynamic metadata. The same master edit serves all three.

Export three copies from the 16 bit or 32 bit master. Do not re edit for each output. The output transform handles gamut and tone mapping.

Verify each export on a calibrated reference. Check gradients in skies and shadows. Look for banding in the 8 bit SDR version.

If banding appears increase dithering or add subtle grain. The HDR version will look clean because 10 bit has four times the steps.

The Monitor Trap: Why What You See Isn't What You Get

Your monitor lies to you. Most consumer panels are 8 bit with frame rate control dithering. They flash two colors rapidly to fake a third.

You see smooth gradients on screen. The file actually has bands. True 10 bit panels display 1,024 levels natively.

No temporal trickery. Professional references like Eizo ColorEdge or Apple XDR show real 10 bit or 12 bit. If you edit on an 8 bit FRC screen you cannot trust shadow gradients.

Enable GPU 10 bit output in Windows or macOS. Use DisplayPort 1.4 or HDMI 2.1. Verify with a test gradient ramp.

If you see steps on the ramp your pipeline is 8 bit somewhere. Fix the weak link before you grade.

Common Mistakes That Turn 14-Bit RAW Into 8-Bit Garbage

Shooting RAW then exporting JPEG from camera. The camera bakes 8 bit with no dithering. Editing in Photoshop 8 bit mode.

The default still exists for legacy plugins. Applying LUTs in an 8 bit timeline. The LUT math quantizes hard.

Rendering intermediate 8 bit proxies for speed. The proxy becomes the master by accident. Saving incremental JPEGs during client review.

Each save adds quantization noise. Using 8 bit monitors without knowing. You approve banded work.

Converting ProPhoto to sRGB before editing. You clip gamut and lose steps. The fix is simple.

Keep the chain high bit depth until final export. Check every handoff point.

Quick Reference: Bit Depth Decision Matrix

Frequently Asked Questions

Does bit depth affect file size?

Yes. Each bit doubles the data per channel. A 16 bit TIFF is roughly twice the size of an 8 bit TIFF.

ProRes 4444 at 12 bit is about double ProRes 422 HQ at 10 bit. Storage is cheap. Editing headroom is not.

Can I fix banding after it appears?

Not completely. You can add noise or dithering to mask it. You cannot restore tonal steps that were discarded.

Prevention at capture and working stages is the only real solution.

Is 32 bit float overkill for stills?

For most stills 16 bit is enough. 32 bit float helps with extreme HDR merges or heavy compositing where intermediate values exceed white point. Photoshop supports it. Lightroom does not need it.

Do I need a 10 bit monitor for 8 bit delivery?

You need a 10 bit monitor to see what the 8 bit export will actually look like. Without it you approve banding you cannot see. Calibrate the display.

Verify the pipeline.

What is the minimum bit depth for HDR video?

Ten bit is the absolute minimum per SMPTE ST 2084. Twelve bit is preferred for mastering. Eight bit HDR does not exist in any professional spec.