ghalfacree - Network

Now that April Fool's morning is well and truly over, I'm pleased to confirm that @rc2014 isn't pulling your leg - well, not...

Now that April Fool's morning is well and truly over, I'm pleased to confirm that @rc2014 isn't pulling your leg - well, not about the existence of the #RC2014 Mini II Picasso, at least.

https://oldbytes.space/@rc2014/114261520883071259

How do I know? Here's my hands-on review of the thing. Spoiler: it's *lovely*.

https://www.hackster.io/news/science-art-and-nostalgia-combined-hands-on-with-the-rc2014-mini-ii-picasso-8686eb339d59

#VintageComputing #RetroComputing #Electronics #Technology #Art

Honestly, sometimes I marvel at the things I get paid to do...#RetroComputing #Z80 #Programming

Honestly, sometimes I marvel at the things I get paid to do...

#RetroComputing #Z80 #Programming

If this all works - and it's a big if - I'll have not only dramatically sped up the whole process, but the finished result will be...

If this all works - and it's a big if - I'll have not only dramatically sped up the whole process, but the finished result will be better with less page content missing from those annoying catalogue pages that don't respect the gutter.

Cross your fingers for me!

That should speed up the scanning stage (fewer rescans due to misfeeds, faster scanning) and the processing stage (no, or at least less,...

That should speed up the scanning stage (fewer rescans due to misfeeds, faster scanning) and the processing stage (no, or at least less, manual deskewing).

But what about the cutting stage? That's a very manual process right now, involving a Stanley knife and a steel rule.

The answer: a heavy-duty guillotine, which the manufacturer claims can do a 4cm-thick stack of paper. That's at least half again as thick as the biggest issue I've got - and it'll cut perfectly straight in one pass (I hope!)

Right, after the disappointment of that last scanner arriving broken (and filthy), it's time to get serious.I've found a more...

Right, after the disappointment of that last scanner arriving broken (and filthy), it's time to get serious.

I've found a more reputable seller and ordered a Fujitsu Fi-7160. No flatbed, I'll keep my current SP-1425 for that, but it's got intelligent anti-skew, supposedly top-end paper feeding tech, and does 120 images a minute(!) from the 50 IPM of my SP-1425.

Fully refurbished, new rollers, 12-month RTB warranty. Should be here some time next week.

Finally, after ordering it on the 6th, got my second-hand document scanner.And it's broken.And filthy.And the power cable's......

Finally, after ordering it on the 6th, got my second-hand document scanner.

And it's broken.

And filthy.

And the power cable's... melted?

"Good used condition" my backside. Now to work on getting my money back...

Had just scanned the last page of a magazine... when the desktop crashed. 300-odd pages.Rebooted... Simple Scan had autosaved! But it...

Had just scanned the last page of a magazine... when the desktop crashed. 300-odd pages.

Rebooted... Simple Scan had autosaved! But it crashed when I scrolled to the end. Never seen it do that before.

Reopened it, told it to export the files. It got to Page 302 before crashing again, which was better than I could have hoped - I only had to rescan 20 pages or so. Phew!

Product photography morning today. Well, early afternoon, too. Always takes longer than you think.Microchip Curiosity Kit, which looks so...

Product photography morning today. Well, early afternoon, too. Always takes longer than you think.

Microchip Curiosity Kit, which looks so similar to the Discovery Kit I thought they'd sent me the wrong thing at first. Different chip at the heart, though: a PIC64 microprocessor.

#photography #technology

Oh, and don't confuse JPEG XL with JPEG XS, which is a different thing.It's easy to remember: JPEG XL creates JPEG files which are...

Oh, and don't confuse JPEG XL with JPEG XS, which is a different thing.

It's easy to remember: JPEG XL creates JPEG files which are Xtra Small; JPEG XS creates JPEG files which are Xtra Large. Easy!

(It actually refers to the size/complexity of the implementation: JPEG XS is designed for resource-constrained devices like microcontrollers and delivers less-well-compressed and thus larger output, JPEG XL is for big heavy beasts like smartphones, desktops, and laptops.)

There's one final JPEG-associated format to try: JPEG XL. JPEG XL is another next-generation would-be JPEG replacement, which offers a...

There's one final JPEG-associated format to try: JPEG XL. JPEG XL is another next-generation would-be JPEG replacement, which offers a range of improvements including optional lossless compression (15.3MB PNG to 9.4MB JXL in four seconds? Yes please!) and some tech inherited from the Fractal Image Format (remember that?)

It's ace. Here I'm using 75% quality, and the file is smaller than any other format we've tried at the same 300dpi. I love it!

...but it's not supported in PDFs yet. Bah!

But you don't have to go to a whole new format to make gains: you can switch to a different JPEG encoder. Google has jpegli: does it...

But you don't have to go to a whole new format to make gains: you can switch to a different JPEG encoder. Google has jpegli: does it deliver improvements?

Yes! The artefacts are noticeably reduced! More texture is retained! And the files are smaller, even! Red's gone a bit pale, tho'.

It also has a shiny new XYB colour space mode, which improves the quality even more - and reduces the filesize! And the reds look better!

But no PDF viewer supports XYB. Most crash; here's MuPDF doing its best.

But JPEG's old tech. We've got new stuff now - like JPEG 2000! A next-gen successor to JPEG, it's patent-encumbered but free...

But JPEG's old tech. We've got new stuff now - like JPEG 2000! A next-gen successor to JPEG, it's patent-encumbered but free implementations are out there.

What does that do? Well, supposedly it delivers a higher quality picture for the same filesize - or a smaller filesize for the same quality. I've read studies, it's proven by Science.

It's got all kinds of whizz-bang features, but the quality difference for this application seems negligible.

One plus: it's supported in PDFs!

Today (well, last night, mostly), I have been investigating JPEG and JPEG-adjacent file formats. (Yes, I'm still tweaking my mag-scan...

Today (well, last night, mostly), I have been investigating JPEG and JPEG-adjacent file formats. (Yes, I'm still tweaking my mag-scan PDF outputs.)

Let's take a look at a baseline. The first image is a PNG, straight from the scanner at 300dpi. I'm pretty sure it's transferred across the wire as JPEG, but lightly compressed. It's the best I can get.

The next one: standard JPEG, 65% quality. That's what my PDFs have now.

The third: standard JPEG, 85% quality but resampled to 150dpi. Nasty.

I could always just resample the entire PDF, but that's an approach of last resort: that way I'm taking JPEGs and recompressing them...

I could always just resample the entire PDF, but that's an approach of last resort: that way I'm taking JPEGs and recompressing them to JPEG again, which ain't ideal.

Although I'm also going from 300dpi to 96dpi, so I guess it wouldn't make *that* much difference to the final quality.

Or I could go the other way and leave the text layer alone but replace the images in the high-res PDF with their lower-res equivalents....

Or I could go the other way and leave the text layer alone but replace the images in the high-res PDF with their lower-res equivalents. Which... I also don't know how to do, without also deleting the corresponding text layer.

Ideas (that don't involve manually editing 700+ pages per mag) welcome!

PDF wranglers! I've got a question, and it's... niche, I think.I scan mags to images then use Tesseract OCR to produce a PDF with an...

PDF wranglers! I've got a question, and it's... niche, I think.

I scan mags to images then use Tesseract OCR to produce a PDF with an invisible but searchable-and-selectable text layer. Works great.

However, I want to produce some low-resolution PDFs. I can do this fine, but the lower resolution image results in terrible OCR performance.

My thought for a fix: extract the text layer from the high-res PDF, apply to the low-res PDF.

But... how? pdftk can't do it, pdftotext only goes one way...

The *quality* is better than I remember, likely because that was a fair few smartphone generations ago. It doesn't even come close to...

The *quality* is better than I remember, likely because that was a fair few smartphone generations ago. It doesn't even come close to what a real scanner can achieve, though - even at 300dpi rather than 600dpi.

These are 1:1 crops, unedited, straight from Genius Scan and a Fujitsu SP-1425 respectively. The Genius Scan example was hand-picked as the best it managed in my brief test; the Fujitsu example picked from a random issue.

The results were better than last time I tried, but still not great: it's not really set up for magazines, but I bet it'd be great...

The results were better than last time I tried, but still not great: it's not really set up for magazines, but I bet it'd be great with wide-margin books or printed paperwork.

Note that the cover is cut off at the top and warped (by a confused *de*-warping algorithm) at the bottom. Left-hand pages have my thumb prominently visible, and reflections - and the gutter is lost, while the page remains warped.

I am still very much on my magazine-scanning kick (https://mastodon.social/@ghalfacree/114069867420929055) - more so, in fact, as I've...

I am still very much on my magazine-scanning kick (https://mastodon.social/@ghalfacree/114069867420929055) - more so, in fact, as I've now taken possession of a bunch of '90s/'00s PC Pro mags I'm currently scanning.

I had someone ask if I'd tried Genius Scan, or a similar smartphone scanner app, to save myself a lot of time. I had, but not for a few years - so I gave it a go!

Latest Genius Scan, configured for "Highest Quality" and dewarping turned on, installed on a Motorola Edge Neo 50. PC Pro, still bound, natural light.

Might have to give up on @Vivaldi thanks to a bizarre bug: the scrollbar is thinner than it seems.Left-most bar in the attached image has...

Might have to give up on @Vivaldi thanks to a bizarre bug: the scrollbar is thinner than it seems.

Left-most bar in the attached image has the mouse cursor five pixels from the right-hand edge, works fine. The bar next to it has the cursor *four* pixels from the edge... and the bar is deactivated. (Without "Native Windows" enabled it also switches to resize!)

Which means you can't swipe your cursor all the way to the right and scroll.

Three-year-old bug, too: https://forum.vivaldi.net/topic/74302/when-the-window-is-not-maximized-only-half-of-the-scrollbar-is-clickable