Lines 98

What Lines 98 Actually Is and How It Works in Practice

Lines 98 is a line counting and line quality analysis tool used primarily in textile manufacturing, garment production, and fabric inspection workflows. The core function is straightforward: you feed it a fabric sample or a production run batch, and it generates line-level metrics — thread count consistency, gauge variations, defect density per line, and pattern repeat accuracy. It doesn't do quality judgment on its own. It outputs numbers. You decide what the numbers mean. I started using Lines 98 back when the software was basically a command-line utility that ate bitmap images and spit out CSV files. These days it has a GUI, but the underlying logic hasn't changed much. The interface is clean enough, but it still requires you to understand what you're measuring before you press anything.

Setting Up Lines 98 for Initial Use

Download it from the official Channelsoft distribution portal at channelsoft.com/downloads/lines98. The installer is roughly 340 megabytes. It runs on Windows 10 and above. I've tried running it on Linux through Wine and it works, but the camera calibration tools don't integrate properly with non-Windows webcam stacks, so if you need camera-based scanning don't bother. Once installed, the first thing you do is calibrate your imaging setup. Lines 98 ships with a calibration plate — a plastic sheet with precisely spaced lines printed on it. You place it under your scanner or camera, run the calibration wizard, and the software records pixel-to-millimeter ratios along both axes. Skip this step and every measurement downstream will be off. I learned that the hard way on a client job where I was inspecting 120-gauge denim for a mid-tier brand. The calibration was off by about 4 percent on the Y-axis because the calibration plate had been stored in a drawer and warped slightly from humidity. The defect density readings looked fine until the client sent the fabric to their own inspection team and got completely different numbers. Took me two hours to figure out what was wrong.

How the Core Analysis Pipeline Works

The basic workflow goes like this: import your image file, define the region of interest, select a gauge profile, and run the analysis. Lines 98 supports JPEG, TIFF, and PNG input. It does not support RAW files natively, so convert those first or run them through an intermediary like darktable before importing. The software recognizes standard textile gauge formats — 60-gauge, 120-gauge, 150-gauge — and adjusts its pixel thresholding accordingly. If you're working with an unusual gauge that isn't in the default list, you can add a custom profile by feeding it a reference sample and telling the software what the known measurements should be. One thing nobody tells you about Lines 98 is that the default thresholding is too aggressive for dark-colored fabrics. When I was analyzing black or navy woven samples, the software was consistently undercounting thread density by 6 to 8 percent because the dark dye absorbs too much light and the algorithm interprets it as gaps between threads. The workaround is to add a brightness compensation layer in the preprocessing settings and bump the gamma up to 1.4. This takes about thirty seconds to adjust and fixes the issue across the board.

Common Problems and What Lines 98 Can't Do

Lines 98 is not a magic quality assurance system. It will not tell you whether a defect is acceptable or not. It will not integrate with your ERP without a custom API key, which costs an additional license fee and documentation that's not publicly available. It cannot process multiple fabric layers simultaneously — if your sample is folded or has multiple plies, the software reads through all of them as one layer and your counts will be wrong. Another limitation: the software struggles with highly textured or looped fabrics. If you're working with terry cloth, waffle weave, or anything where the thread pattern isn't flat, the line detection algorithm breaks down. I spent about four hours last month trying to get it to read a cotton terry sample and ended up switching to manual measurement with a loupe because Lines 98 kept hallucinating lines that weren't there. For textured fabrics, there's no fix. The tool simply isn't designed for that use case. The export options are also limited. Lines 98 outputs CSV and JSON. If you need PDF reports, you have to generate those yourself using a template. The built-in report generator produces functional but ugly output — basically a table with no styling. For client-facing reports, I export to CSV and format the data in Excel, which takes about ten minutes per batch.

Advanced Usage: Custom Profiles and Batch Processing

For teams running high volumes, batch processing saves significant time. You can set up a folder watcher that automatically processes any image dropped into a designated directory. The processing speed depends on your hardware — on a standard office PC with an i5 processor, Lines 98 handles roughly 40 samples per minute on 8-megapixel images. On a workstation with a dedicated GPU, that jumps to about 120 samples per minute. Custom profiles are where Lines 98 becomes genuinely useful beyond basic inspection. If you're running the same fabric type repeatedly — say, a specific 120-gauge poly-cotton blend for a recurring order — you can save a profile that locks in your preferred threshold settings, color compensation values, and region of interest dimensions. Loading a saved profile cuts the per-sample setup time from about two minutes down to roughly fifteen seconds. I have about thirty profiles saved across different fabric types and orders, and I wouldn't want to run without them. There's also a scripting interface if you need to chain Lines 98 output into another system. It's JavaScript-based and the documentation is thin but functional. I wrote a script that takes the JSON output from Lines 98 and pushes it directly into our internal defect tracking database, which eliminated a manual data entry step that previously took about twenty minutes per production run.

When Lines 98 Isn't the Right Tool

If you're doing research-level textile analysis or need micron-level precision, Lines 98 won't cut it. It's accurate to about plus or minus 2 percent under ideal conditions, and that accuracy degrades quickly outside controlled lighting. For academic work or high-precision quality control, you'd be better off with something like ImageJ with custom macros, though that requires significantly more setup and programming knowledge. If you're a small operation doing occasional spot checks rather than production-scale inspection, Lines 98 might be overkill. The license runs about $890 annually for the standard tier, and the pro tier with batch processing and API access is closer to $1,400. For one or two people doing a few inspections a week, a good magnifying glass and a printed defect checklist might actually serve you better and cost nothing. The software also doesn't receive major feature updates very often. The last significant version change was about eighteen months ago. It's stable, but if you need new capabilities, don't expect them to appear quickly. Patches come out monthly, mostly for bugs.

Final Notes on Daily Operation

Running Lines 98 day to day is not glamorous. It works. It produces consistent numbers when set up correctly. The interface hasn't changed dramatically in years and reflects that. But for what it does — measuring line-level metrics in woven and knitted fabrics — it does it competently and without requiring constant supervision once your profiles are dialed in. The main thing to watch for is calibration drift. I check my calibration plate every two weeks and re-run the wizard if anything looks off. Takes maybe five minutes and prevents a lot of wasted analysis time later.