The latest articles on DEV Community by Damon Charlie (@damoncharlie). https://dev.to/damoncharlie https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4178600%2Ff75ecae7-61f7-4def-8e37-cafcff53887f.png https://dev.to/damoncharlie en Damon Charlie Mon, 12 Oct 2026 07:06:57 +0000 https://dev.to/damoncharlie/converting-office-files-to-markdown-and-back-with-markany-244g https://dev.to/damoncharlie/converting-office-files-to-markdown-and-back-with-markany-244g <p>I keep notes in Markdown and receive everything else: PDF, Word, Excel, PowerPoint, HTML, EPUB, and photos of whiteboards. Retyping those into headings and lists is the tax I used to pay before editing. MarkAny batch-converts those formats into structured Markdown, and it can export Markdown back out to common formats. It also runs text recognition, so a picture of a page is not a dead end.</p> <h2> What I convert, and why </h2> <ul> <li>A PDF spec becomes headings and code fences I can diff</li> <li>A slide deck becomes one section per slide, which is easier to edit than text boxes</li> <li>A spreadsheet becomes a table I can review in a plain-text diff</li> <li>An EPUB or an HTML page becomes a single note</li> <li>A photo of a whiteboard becomes text I can correct, instead of a file I cannot search</li> </ul> <h2> A batch habit </h2> <p>I do not convert one file at a time if I have a folder. I drop the set in, let the batch finish, then open the Markdown and fix the places text recognition is weak: columns that merged, headers that became body text, and tables that lost a cell.</p> <h2> Going back out </h2> <p>Once the note is clean I sometimes need a Word file or a slide again for someone who will not open Markdown. Exporting from the cleaned source is better than editing the original binary. The Markdown stays the copy I trust.</p> <h2> Limits I plan for </h2> <p>Text recognition on handwriting is a draft, not a record. I read every line that came from a photo. Scanned tables with merged cells often need a manual rebuild. MarkAny removes the retyping. It does not remove the proofread, and for anything I will publish, the proofread is the whole point.</p> ai productivity tutorial webdev Damon Charlie Mon, 12 Oct 2026 07:06:41 +0000 https://dev.to/damoncharlie/translating-files-while-keeping-the-structure-with-transmonkey-5gfj https://dev.to/damoncharlie/translating-files-while-keeping-the-structure-with-transmonkey-5gfj <p>Most translation tools are fine for a paragraph and clumsy for a file. A slide deck comes back as a wall of text. A scanned form loses its boxes. A video comes back as a transcript with no timing. TransMonkey is an AI translator aimed at those three cases: documents, images, and video, with the structure of the original kept intact.</p> <h2> Documents </h2> <p>For a report or a worksheet I want headings, lists, and tables to survive. My check is simple: open the source and the translation side by side and confirm that section 3 is still section 3. If a table column swapped meaning, I fix that cell by hand instead of re-running the whole file.</p> <h2> Images </h2> <p>Photos of signs, slides, and screenshots need the words translated where they sit, not dumped into a separate caption. I use this for UI screenshots when a teammate needs to read a menu that is not in their language. I still zoom in. Small text is where text recognition guesses.</p> <h2> Video </h2> <p>For a short clip I want a translation I can read in time with the picture. I watch once with the translation and once without, and I note any line where a joke or a product name was translated too literally. Names of tools and people should usually stay as they are.</p> <h2> A habit that prevents bad publishes </h2> <p>I never ship a translation I have not sampled. For a long document I read the introduction, one middle section, and the conclusion. For a video I spot-check the first minute and a minute in the middle. TransMonkey gets the structure across formats. The accuracy pass is still mine, especially for numbers, names, and anything a reader might act on.</p> ai productivity tutorial webdev Damon Charlie Mon, 12 Oct 2026 07:00:03 +0000 https://dev.to/damoncharlie/drafting-quizzes-and-lesson-plans-with-teachany-fig https://dev.to/damoncharlie/drafting-quizzes-and-lesson-plans-with-teachany-fig <p>Blank-page planning is the slow part of teaching prep. The ideas are usually there. What takes the evening is turning them into a quiz, a rubric, and a lesson plan that a substitute could follow. TeachAny is a set of AI tools aimed at that paperwork: quizzes, rubrics, lesson plans, and other classroom materials.</p> <h2> How I use it without handing over the class </h2> <p>I treat the tool as a first draft, not as the teacher.</p> <ol> <li>Write the learning goal in one sentence before opening anything. Example: students can explain why a loop stops.</li> <li>Ask for a short quiz that checks that goal, with a mix of recall and one applied question.</li> <li>Read every item. Delete anything that tests trivia the lesson never taught.</li> <li>Generate a rubric only after the quiz is stable, so the criteria match the actual task.</li> <li>Build the lesson plan last: opener, the worked example, practice, and a check for understanding.</li> </ol> <h2> What to keep an eye on </h2> <ul> <li>Answer keys can be confidently wrong. I solve the quiz myself before I print it.</li> <li>Reading level drifts. A prompt for a middle grade sometimes comes back sounding like a textbook for adults. I rewrite the stems in the words I would say out loud.</li> <li>Rubrics love vague rows like "shows understanding." I replace those with something a student can see, such as "labels the stopping condition of the loop."</li> </ul> <h2> A small set of materials is enough </h2> <p>For one lesson I usually leave with four things: a five-question quiz, an answer key I verified, a four-row rubric, and a one-page plan. TeachAny gets me to that pile quickly. The part I do not skip is the edit. Students notice a wrong key faster than any other mistake in the packet.</p> ai productivity tutorial beginners Damon Charlie Mon, 12 Oct 2026 06:59:07 +0000 https://dev.to/damoncharlie/a-browser-workflow-for-cleaning-up-photos-with-imgkits-2l4j https://dev.to/damoncharlie/a-browser-workflow-for-cleaning-up-photos-with-imgkits-2l4j <p>I keep running into the same photo jobs: a stray object in the corner, a background that fights the subject, and a scan that looks dull. Opening a full desktop editor for each one is slower than the job itself. Lately I have been doing those fixes in the browser with Imgkits, an all-in-one AI image and video editor.</p> <h2> What it covers </h2> <p>Imgkits is built for the small, repetitive edits that show up in content work:</p> <ul> <li>Removing an object or a person who walked into the frame</li> <li>Restoring a photo that is faded, scratched, or low resolution</li> <li>Replacing or simplifying a background</li> <li>Enhancing clarity and applying a light creative filter</li> <li>The same kind of edits on short video clips, not only stills</li> </ul> <h2> A workflow that stays short </h2> <ol> <li>Start with the original file. Do not upscale first. Cleanup tools behave better on the real pixels.</li> <li>Remove the distraction before you touch color. Object removal is easier to judge on an untouched image.</li> <li>Restore or enhance second. If the file is a scan or an old phone photo, this is the step that recovers edges.</li> <li>Edit the background last. Once the subject is clean, a simpler background reads more clearly.</li> <li>Export and compare against the original at the same zoom. If a repaired edge looks smeared, undo that one change rather than stacking another filter.</li> </ol> <h2> When this is the wrong tool </h2> <p>If you need precise layer masks, color-managed print output, or frame-by-frame animation, a desktop suite is still the better home. Imgkits is the faster path when the job is "make this asset usable" and you want to stay in the browser.</p> <p>I have been using it for product screenshots and family scans. The useful part is not a single magic button. It is having removal, restore, background, and enhance in one place so you do not bounce between four tools for one photo.</p> ai webdev productivity tutorial