Batch convert photos faster: what changed in Contenta Converter 9.0.31

Published September 23, 2026 · 6 min read

Contenta Converter 9.0.31 converts the same photos with less work. Exporting a JPEG takes about a third less processor time than in 9.0.30 and exporting a PNG about a quarter less, and the picture comes out the same. Together with 9.0.30, released the day before, a batch of 48 sharpened JPEGs that took 12.8 seconds now takes 9.0 seconds on a 4-core PC.

Every number on this page was measured on real photos on one PC, our development machine, limited to 4 processor cores because 4 cores is the most common processor among Contenta Converter users. Your times will differ with your processor and disk. The ratios are what carry over.

Before and after, measured

Measured September 20 to 22, 2026 on one PC, limited to 4 cores (single-file rows: one conversion process), median of 3 to 12 repeated runs. Before = the previous version of the same step. Processor time is the work the computer does; a laptop on battery or a PC busy with other programs feels it most.

What this means for 500 photos

The 48-photo sharpen batch works out at 0.27 seconds per photo before and 0.19 seconds after. For 500 photos of the same kind (camera JPEGs of 2 to 10 megapixels, quality 90, sharpen) that is about 2 minutes 14 seconds before and about 1 minute 34 seconds after, on 4 cores. The converted files also take about 0.6 GB instead of 1.8 GB, because sharpened photos are no longer saved at quality 100.

This estimate scales the measured 48-photo batch in a straight line and only counts the changes in 9.0.30. The faster JPEG encoder in 9.0.31 comes on top: it cut the processor time of the same kind of batch by another 14%, but the test PC was running other work during that measurement, so we do not quote a clock time for it.

What changed

A faster JPEG encoder, same pictures

JPEG files are now encoded with libjpeg-turbo, which uses the SIMD instructions of modern processors. We kept the previous encoder's colour handling and file layout, so the files look and measure the same: across 201 test images the median quality score (PSNR) moved by 0.02 dB at most, the median file size by less than 0.1%, and the metadata (EXIF, XMP, IPTC, colour profile) is copied as before. One photo needs 34% less processor time; a 48-photo batch 22% less.

Sharpened photos keep your quality setting

Before 9.0.30, any effect (sharpen, colours, sepia and the others) made the app save the JPEG at quality 100, whatever the quality setting said. Sharpen is the most-used operation in Contenta Converter, so many exports came out three to four times larger than needed. A 10-megapixel photo exported at quality 90 with sharpen was 4.75 MB; it is now 1.22 MB. The sharpen filter itself also runs about three times faster, with the same result pixel for pixel.

Each file is written once

Contenta Converter keeps your photos' metadata (camera, date, GPS, copyright) by default. Before, each converted file was written to disk, read back and written a second time to add that metadata. Now the metadata is added in memory and every JPEG and PNG is written once, which halves the data written. On a fast internal SSD you will barely notice. On a USB stick, an external hard drive or a network share, the second write was the slow part.

PNG: faster by default, and the compression setting now works

PNG exports now use compression level 2 by default and need about a quarter less processor time. The trade-off is size: our test photo came out at 12.4 MB instead of 11.6 MB. The PNG compression setting in the app and on the command line used to be ignored by the converter; now it is applied. Set it towards 9 for the smallest files or to 0 for the fastest. The picture is identical at every level, because PNG is lossless.

Camera RAW files use every core

RAW decoding (Canon CR2, Sony ARW, Nikon NEF and the other RAW formats) used to run on a single processor core. It now uses all of them. The demosaic step went from 2.0 to 0.63 seconds for a Canon CR2 on 4 cores, from 1.68 to 0.53 seconds for a Sony ARW and from 1.15 to 0.43 seconds for a Nikon NEF. The gain is largest when you convert one RAW file or a few: in a large batch every core is already busy with its own file, so a big batch gains much less. The decoded pictures are identical: we compared every RAW file in our test set, byte for byte.

Resizing for the web on NVIDIA PCs

On a PC with an NVIDIA graphics card, the app sent JPEG resizing to the graphics card. For photos reduced to half size or smaller the processor turned out faster, because it can decode a JPEG directly at a reduced size: 48 photos resized to 1920 pixels took 10.6 seconds on an RTX 3080 and take 5.9 seconds on the processor (4 cores). These jobs now stay on the processor. Full-size JPEG exports still use the graphics card, where it wins.

The command line starts in half the time

The contenta command line checked the licence online before every job, which cost 0.7 to 1.7 seconds. A registered copy now reuses a check that is less than 24 hours old, so a one-file job went from 1.20 to 0.57 seconds. Scripts and AI agents that call contenta convert once per file gain this on every call. Trial copies never made that check, so nothing changes for them. How to automate conversions with the CLI

Fixes you will notice

• Portrait RAW photos are no longer rotated twice. The RAW decoder turned the picture upright, then the original orientation tag was copied onto it, so photo viewers turned it again and showed it sideways. The tag is now reset; camera, lens, GPS and date details are copied as before.
• RAW exposure compensation works. With the default settings (noise reduction on), the exposure compensation value was lost before it reached the RAW decoder, so the setting did nothing. It is now applied.
• Decimal values on European PCs. On Windows set to a decimal comma, as in most of Europe, a value such as 0.5 was read as 0, so fractional effect settings (gamma, colour temperature) and RAW exposure steps were cut off. They now come out the same on every PC.
• Kodak Photo CD files in folders with non-Latin names. A PCD file in a folder named in Japanese, for example, failed with a missing-tool message. It now opens at its full 3072 × 2048 resolution.

How to get 9.0.31

If Contenta Converter is already installed, start it: a banner says an update is available and links to the download page. Run the new installer over the old version; your settings and licence stay in place.

New to Contenta Converter? The free trial converts 10 images without a watermark, so you can time your own photos on your own PC before you buy. The licence is $129 once, with no subscription.

Questions