Adobe vs ArtCraft · PhotoCraft, FilmCraft and EffectCraft Tested

Quick Summary

• AI Samson compares matched tasks using the same source files and prompts. His rubric separates output quality, elapsed time and manual usability, which prevents a fast agent command from being mistaken for a complete professional replacement.
• The headline references seven Adobe-style apps, but the direct timed tests focus on three pairs: Photoshop versus PhotoCraft, Premiere Pro versus FilmCraft, and After Effects versus EffectCraft. VectorCraft, LightCraft, PdfCraft and DesignCraft belong to the wider suite but do not receive equivalent hands-on tests in this video.
• PhotoCraft wins the measured speed comparisons for a 20-image badge batch, face blurring and social-media resizing. It matches Photoshop on these controlled jobs, but Photoshop wins the more demanding person composite and object-removal tests.
• PhotoCraft completes the 20-image badge task in under 40 seconds versus 1 minute 23 seconds for Photoshop. Face blur takes 24 seconds versus 30, and five social crops take 28 seconds versus 1 minute 14. These are creator-measured results from one setup, not universal benchmarks.
• FilmCraft can create a transcript-guided talking-head cut, assemble a precisely specified montage and generate a captioned vertical short. Its difficult audio cleanup remains weaker, it lacks ready-to-use text-based editing in release builds, and the montage export takes about two minutes versus eight seconds in Premiere Pro.
• Both editors make mistakes on the automated short. Premiere produces awkward caption styling and weak reframing. FilmCraft produces cleaner-looking captions but worse timing, mistranscribes 80 as 18, misses the requested Poppins font and also fails to reframe the important visual content.
• EffectCraft is the strongest surprise. The lower third, kinetic-text quote and logo-reveal tests are judged close enough to draws, although Samson notes a distracting bounce in the EffectCraft lower third. Both lower-third builds take about 48 seconds in his setup.
• The conclusion is selective adoption, not cancellation. Use the Crafting Apps for bounded, reversible tasks and agent-driven experiments. Keep mature software for demanding audio, compatibility, plug-ins, client handoff, large projects and deadlines until your own benchmark proves otherwise.

Credit, publication dates and what was actually tested

The embedded video was created by AI Samson. YouTube records its source publication at October 14, 2026 at 11:53 p.m. Pacific time, which was October 15 in Lisbon. This JQ VISIONS article was published on October 16, 2026 and uses evidence checked through that date.

Samson says he gave each pair the same files and the same prompt, then judged result quality, completion time and manual usability. That is a valuable creator test, but it is not a laboratory benchmark. Machine, operating system, application version, MCP behavior, caching and Claude variability can all change the result.

The seven original Adobe-style Crafting Apps target Photoshop, Illustrator, Premiere Pro, Lightroom, Acrobat, After Effects and InDesign workflows. The detailed comparisons here cover only PhotoCraft, FilmCraft and EffectCraft. The other apps are part of ArtCraft's wider lineup, not winners or losers in this specific scorecard.

What ArtCraft is in October 2026

ArtCraft's official directory now lists thirteen Crafting Apps, not only the original seven. The project describes them as native Rust applications, open source under permissive licenses, local-first and agent-ready through command-line, JSON control and MCP interfaces.

PhotoCraft and PdfCraft are labeled early alpha on the official directory. FilmCraft and EffectCraft are labeled in development. Those labels matter more than social-media shorthand such as Adobe replacement. Early software can complete impressive demonstrations while still having gaps in compatibility, reliability, accessibility, performance and support.

Agent-ready architecture is the genuine new capability. Claude can issue structured commands to the project model instead of moving a cursor through every menu. That helps explain the speed of simple batch tasks, but it does not improve an underlying effect, codec, font engine or repair algorithm beyond what the app implements.

The three-part rubric: quality, time and usability

Quality asks whether the exported asset is visually or sonically acceptable. Time measures the complete requested task, not merely how quickly the agent begins. Usability asks whether a person can enter the app, adjust the result and recover when automation fails.

These scores can disagree. PhotoCraft is faster on several batches, yet Photoshop produces the better composite. FilmCraft matches the edit structure of a montage, yet its export takes much longer. EffectCraft generates persuasive motion graphics, while its official gap list still says existing After Effects projects and plug-ins cannot be opened.

A fair migration test should add two more dimensions: interoperability and reliability. Reopen the native project, export it twice, inspect the result in another tool and repeat the task on a second machine. A successful demo is not equivalent to a dependable production pipeline.

PhotoCraft test 1 · Twenty branded images

The first brief asks each app to open twenty photos, size a yellow AI Samson badge for each image, place it at bottom right and export every result. Both tools position the mark correctly. PhotoCraft completes the batch in under 40 seconds, while Photoshop takes 1 minute 23 seconds in Samson's recording.

The advantage comes from direct structured commands. PhotoCraft was designed to be driven by an agent, so the operation can be expressed as one repeatable batch rather than twenty visible interactions. Photoshop still completes the task quickly, but its automation path in this setup is less direct.

This is an ideal adoption candidate because the job is deterministic, reversible and easy to inspect. Before using it on paid assets, add file naming, color-profile preservation, overwrite protection and a manifest confirming that all twenty outputs match the requested dimensions.

PhotoCraft tests 2 and 3 · Face blur and social crops

The face-blur brief asks for two people to be made unrecognizable with soft edges. PhotoCraft finishes in 24 seconds and Photoshop in 30 seconds. Samson considers the exports almost indistinguishable and scores the task as a draw.

The resizing brief creates five platform versions from one photo, including YouTube, Instagram, an Instagram Story and LinkedIn. PhotoCraft takes 28 seconds while Photoshop takes 1 minute 14. The visible outputs are again judged effectively identical.

Do not convert this into a blanket claim that PhotoCraft is twice as fast. These are two short tasks on one setup. The supported conclusion is narrower: simple, explicit transformations are where a structured agent interface can remove repetitive interaction and produce a meaningful speed advantage.

PhotoCraft tests 4 and 5 · Composite and object removal

The composite brief cuts Samson out of one camera photo, places him on the right side of a stage image and matches the room color. Both tools finish in roughly the same time, with Photoshop slightly faster. Photoshop retains more attractive lighting and earns the win on output quality.

The removal brief deletes a sign and two price tags while reconstructing the shelf. The automated PhotoCraft attempt fails, although Samson gets a much better result when he paints over the objects manually with its remover. Photoshop produces the stronger automated repair, but its reconstructed background is not perfectly logical either.

This is the central PhotoCraft lesson. Feature presence and agent control do not guarantee intelligent reconstruction. The app is credible for occasional image work, while complex selection, compositing and generative repair still favor Photoshop in this test.

PhotoCraft verdict · Fast for bounded work, not a full Photoshop replacement

PhotoCraft wins or draws the simple jobs and loses the two demanding image-editing tasks. Samson's practical conclusion is that Photoshop remains superior for professional work, while PhotoCraft can be useful when the subscription cost is the deciding constraint.

The official PhotoCraft project supports real PSD files, layers, masks, type, vectors and adjustment workflows, but its own gap document continues to record missing formats, hardware limitations, localization gaps and incomplete measured fidelity. Those are normal early-alpha conditions, not minor footnotes.

Use a round-trip test before trusting a layered file. Open a copied PSD, save under a new name, close it, reopen it in PhotoCraft and then inspect it in Photoshop or another trusted reader. Check layer order, text, masks, blend modes, color profile, bit depth, smart objects and exported pixels.

FilmCraft test 1 · Difficult audio cleanup

The audio clip contains fan noise and electrical hum. The brief asks for clear speech, YouTube-ready levels and a WAV export. Both automated results struggle, but Premiere Pro becomes substantially better when Samson adjusts its enhancement controls manually.

FilmCraft improves the recording but not enough for him to use it. On a separate higher-quality source that needed only modest enhancement, he reports much better results. This distinguishes light polish from restoration: a basic denoiser can help an already usable recording without rescuing severely damaged sound.

For production, keep the raw recording, measure loudness and true peak, listen on headphones and speakers, and compare the processed file for metallic artifacts or lost consonants. If intelligibility depends on the repair, use the most reliable audio tool available rather than treating free as the primary criterion.

FilmCraft test 2 · Removing retakes from raw footage

Samson supplies four and a half minutes of retakes, coughs and false starts. He asks Claude to read the transcript, keep the clean takes, create one tight version and export it. Both FilmCraft and Premiere Pro return the same final duration and both clip a few words too aggressively.

FilmCraft does not expose the same ready text-based editing workflow in the release Samson tests. Its official gap document explains that speech-to-text is disabled in release builds, so Transcribe Sequence, text-based editing and automatic captions currently require a source build with the Whisper feature.

The agent can still use transcript information to assemble a cut, which is impressive but not equivalent to a mature editable transcript panel. Review every edit with handles before delivery. A cut that removes the beginning or end of a word should fail the rubric even if the total duration looks correct.

FilmCraft test 3 · A precisely specified montage

The montage uses ten clips at two and a half seconds each, cross dissolves, fade in, fade out and one music track. Both editors follow the structured brief successfully, so Samson scores output quality as a draw.

Export performance is not close. Premiere Pro takes roughly eight seconds while FilmCraft takes about two minutes on his machine. A correct timeline can still be the wrong operational choice when a team needs repeated previews, multiple aspect ratios or a deadline export.

This is the best FilmCraft use case in the video: explicit durations, named transitions, fixed assets and a clear deliverable. It is also easy to validate programmatically by checking sequence duration, cut points, dissolve lengths, audio presence, frame count and final black frames.

FilmCraft test 4 · An automated captioned short

The most ambitious brief asks each editor to transcribe a source video, choose the strongest 30 to 45 seconds, create a vertical short, add burned-in Poppins captions with black text on yellow and export it.

Premiere's output uses the requested font but adds an unattractive outline and fails to reframe the most interesting content. FilmCraft's captions look cleaner, but their timing is worse, the transcript turns 80 into 18, Poppins is not used and the visual framing is also weak. Samson scores the overall task as a draw because neither version is publication-ready.

This test combines editorial selection, speech recognition, typography, font availability, timing, reframing and export. One prompt hides at least six separate acceptance tests. The fix is to approve the transcript and crop before rendering captions, then review the result at phone size with sound.

FilmCraft verdict · Capable assembly, unfinished migration path

FilmCraft handles basic cuts and a clearly specified montage. Premiere Pro wins difficult audio restoration and export speed, while both tools need correction on the captioned short. Samson does not recommend replacing Premiere Pro outright.

The official FilmCraft gap list is unusually candid. It identifies native .prproj import, partial hardware acceleration, real-world media testing, plug-in hosting, delivery formats, platform reliability and text features in release builds as unfinished areas. FilmCraft uses its own readable .fcproj project format and supports interchange formats rather than native Premiere projects.

A safe pilot uses camera files from your actual production, not only clean test clips. Import, cut, grade, mix, title and export. Reopen the project on another machine, verify synchronization and compare delivery files. Keep Premiere, Resolve or another proven editor available until the whole path passes.

The Artlist segment is sponsored and demonstrates an ecosystem advantage

The video's Artlist section is explicitly sponsored. Samson demonstrates image, video, voice and music generation inside Adobe applications, including creating an abstract background in After Effects and adding the downloaded result directly to the project.

The useful observation is ecosystem depth, not an independent product verdict. Mature Adobe applications have decades of plug-ins, templates, integrations, training and professional handoff conventions. FilmCraft's official gap list says plug-in hosting is not yet available, and EffectCraft cannot run After Effects plug-ins.

A new open-source app does not need to reproduce every legacy integration to be valuable. It does need to support the few integrations your workflow depends on. List those dependencies before migration, including fonts, codecs, panels, stock libraries, color tools, caption systems and delivery services.

EffectCraft test 1 · Animated lower third

The brief asks for AI Samson in Poppins black on white, a yellow highlight bar that wipes on, a hold and an animated exit. Both applications take about 48 seconds in Samson's setup.

After Effects produces polished easing but shows a small alignment inconsistency. EffectCraft comes close and earns a draw, although Samson notices a distracting bounce before the exit. This is exactly the kind of defect a frame and velocity review should catch before delivery.

Because the project remains editable, the bounce can be treated as a keyframe problem instead of regenerating a flattened clip. That editability is a major reason to use code or agent-driven project files for repeatable branding work.

EffectCraft tests 2 and 3 · Kinetic type and logo reveal

The kinetic-type test asks each line of a five-second quote to punch in sequentially with one keyword highlighted in yellow. Samson considers the After Effects result elegant and the EffectCraft version impressive enough for another draw.

The logo reveal asks After Effects text to appear, receive a strike-through, drop away and reveal the EffectCraft logo with the word free. Both tools follow the requested sequence effectively and again receive a draw.

These are short, graphic, deterministic jobs. They do not test long footage, hundreds of layers, expressions from existing client projects, plug-ins or a 4K production. They do show that EffectCraft can already be useful for original title cards, lower thirds and social motion elements built from a clear brief.

EffectCraft verdict · The most promising result, with major boundaries

EffectCraft is Samson's strongest ArtCraft performer. It draws all three motion-design comparisons and demonstrates that agent-driven keyframes, type and shape animation can produce credible short graphics quickly.

The official project is more cautious. Its October 12 gap assessment estimates readiness for real professional work at about 45 percent. It cannot open .aep, .aepx, .mogrt or .ffx files, cannot run After Effects plug-ins, has limited fidelity measurement and still records stability and platform issues.

Use it for new, self-contained projects rather than migrating a client's After Effects archive. Save source, export a mezzanine or image sequence, inspect every frame around transitions and keep the original app available for templates or plug-in-dependent work.

The final scorecard

PhotoCraft: three simple-task draws or speed wins, followed by two Photoshop wins on complex compositing and removal. Best current fit: occasional editing, batches, social resizes and reversible image work. Main risk: incomplete fidelity and advanced repair.

FilmCraft: a draw on transcript-guided cutting, a structured montage draw with a major export-speed penalty, a Premiere win on difficult audio and an unsatisfactory draw on the automated short. Best current fit: explicit assemblies and experiments. Main risks: project migration, plug-ins, speech features, formats and reliability.

EffectCraft: three creator-scored draws across a lower third, kinetic type and logo reveal. Best current fit: new short-form motion graphics with simple assets. Main risks: existing project import, plug-ins, fidelity, real-world scale and cross-platform stability.

Overall: do not cancel an established professional tool because one test passes. Run your own representative project, count manual cleanup and recovery time, and switch only when the entire delivery path wins repeatedly.

Clean-room claims, legal uncertainty and ethical debate

The ArtCraft repositories describe the apps as clean-room reimplementations written from scratch in Rust, using public specifications, documentation and observed behavior rather than Adobe source code or proprietary assets. That is the project's stated method, not an independent legal finding.

Copyright, contract restrictions, patents, trademarks and interface trade dress can raise different questions, and their treatment varies by country. The video reports that Adobe declined to comment and that the issue could ultimately be tested in court. This article does not decide that question and is not legal advice.

For a user, the practical ethical test is broader than legality. Credit the project, follow its licenses, avoid confusing clients about what software produced a file, do not distribute Adobe assets or cracked software, and verify that every model, plug-in, font and media asset used in a commercial project permits that use.

A seven-day adoption test before changing subscriptions

Day 1: choose three representative tasks and copy the source files. Day 2: record the result in your current tool with time, quality and friction. Days 3 and 4: repeat in the Crafting App, once manually and once through the agent interface. Day 5: reopen, revise and export every project on a second session or machine.

Day 6: have another person compare blinded exports. Score task completion, fidelity, editable structure, speed, stability and handoff. Count setup, corrections, failed runs and render time. Day 7: decide whether the app is ready for experiments, internal work, selected client tasks or no production use yet.

Set a stop rule before testing. If the app crashes repeatedly, changes the source, loses layers, cannot reopen its project or fails a required delivery format, stop and preserve the evidence. Do not move the next paid deadline into an alpha tool to prove a point.

Verdict · Do not replace the suite, replace individual jobs

Samson's tests do not show Adobe being replaced. They show three open-source applications becoming useful much earlier than their age would suggest. PhotoCraft can remove repetitive image work, FilmCraft can assemble explicit edits, and EffectCraft can produce credible original motion graphics.

The strongest strategy is task substitution. Move one bounded job when the free tool meets the rubric, keep the source and fallback, then expand only after repeated success. This captures the value of agent-native software without asking an early project to carry an entire professional pipeline.

EffectCraft is the standout creative result, PhotoCraft is the easiest practical utility and FilmCraft has the largest operational distance to cover. All three are worth watching. None earns an unconditional cancel Adobe recommendation from this evidence.