ImageViewer: Clamp zoom factor to sane values #9529
Conversation
Should avoid egregious values that would potentially alloc insanely large buffers (and likely fail to do so).
My napkin math was based on *total* math, so I'll probably bump the percentage...
What we get by the time _render has finished has most likely already been prescaled...
ImageWidget Otherwise the max factor gets ever so slightly smaller each time because we eat more memory at higher zoom levels ;). Which makes the final zooming steps confusing as it jitters back and forth.
|
I'm still mildly unhappy with how fast it zooms out and how slow it zooms in (but mostly by the zoom out, the zoom in is somewhat okay), but that was admittedly not the focus here ;). Re #9524 |
It's stupid, but it more or less works, maybe?
|
Okay, tried something stupid about the zoom outs. At first glance, it feels slightly less worse. It's stupid, though ;p. |
It just feels random :) Mixed feelings/thoughts about the min/max scale factor depending on free mem:
|
That's the idea ;o).
Again, the idea is that, if we can't reach it, it's because it would very likely have crashed. (and/or been extremely slow because of general memory pressure).
MMU magic ((mostly) irrelevant concern ;)). Hence heuristics entering the stage instead of assuming all available memory is fair game. |
|
Would need a bit of maths (or the mood to get into that):
|
I napkin math'ed something like that, and it felt like it would lead to (usually) large jumps. Haven't tried it in practice, though, and I agree that it sounds a lot saner ;). |
Whelp, a good night's sleep helps. It makes a hell of a lot more sense, and it finally works sensibly :}. |
frontend/ui/widget/imagewidget.lua
Outdated
| function ImageWidget:recomputeScaleFactor(scaling) | ||
| local req_width = self._bb:getWidth() * scaling | ||
| local req_height = self._bb:getHeight() * scaling | ||
|
|
||
| -- Best fit | ||
| return math.min(req_width / self._img_w, req_height / self._img_h) |
There was a problem hiding this comment.
Isn't self._bb:getWidth() == self._img_w (idem for height) ? Aren't A and B the same in the the math.min(A, B) ?
There was a problem hiding this comment.
self._bb is the scaled bb, while self_img_* is the unscaled image (i.e., the actual reference to compute the scale factor).
There was a problem hiding this comment.
i.e., the two only match in an hypopthetical scale_factor == 1 scenario.
There was a problem hiding this comment.
Aren't A and B the same in the the math.min(A, B) ?
They ought to be since we're only concerned with best-fit w/ AR preserved here, yeah.
There was a problem hiding this comment.
OK, but we do not crop (and I'd like this to stay as is : zoom takes time, but panning is fast as it works with the same self._bb).
So, self._bb and the unscaled image keep the same A/R.
So, it looks to me that req_width / self._img_w == req_height / self._img_h == scaling.
(But I may miss something :) I'm just having a look at your diffs...)
There was a problem hiding this comment.
Well, it looks to me that it comes down to:
function() ImageWidget:recomputeScaleFactor(scaling)
return self.scale_factor * scaling
end
(except for the cases where there is not scale_factor or it is nil or 0 :) but we could as well store the scale_factor we use the last time as a property - and may be we are already doing that.)
But it doesn't feel like documentation or defensivness: it feels like there's some clever essential computation here and that one should stop here to understand it (like I'm doing :) and well, I stopped and I'm still stopped :) The comments is still confusing. It feels like someone got lost in maths and is taking too much precautions :) which I understand :) Just sharing so you might spend more time to be sure it's needed.
There was a problem hiding this comment.
function() ImageWidget:recomputeScaleFactor(scaling)
return self.scale_factor * scaling
end
It is very close to that (as in, the only difference introduced is because of minute differences between w & h, the various truncation passes, and rounding errors).
(except for the cases where there is not scale_factor or it is nil or 0 :) but we could as well store the scale_factor we use the last time as a property - and may be we are already doing that.)
It's already re-computed first in IV if it's 0.
I can go back to scale_factor * scaling if you think it's clearer, it won't change much in terms of actual results ;).
There was a problem hiding this comment.
I do think it's clearer - in the sense it doesn't deceive you by making you think it's doing more and different than the previous self.scale_factor = self.scale_factor + inc :) (while I was somehow hoping it did different :)
(ie. it's clearer having this single line in ImageViewer, that having to follow the trail into ImageWidget to somehow conclude it only does this).
There was a problem hiding this comment.
Oh, it is different, since it's actually matching what rescaling the image does (i.e., scale_factor = scale_factor * (1 + inc) or * (1 - dec), which is different than what happened previously, where we just uniformly added/took a chunk of the factor directly.
There was a problem hiding this comment.
Thanks. The change indeed did not jump as me, like it does now :)
frontend/ui/widget/imagewidget.lua
Outdated
| local req_height = self._bb:getHeight() * scaling | ||
| logger.dbg("From", self._bb:getWidth(), self._bb:getHeight(), "to", req_width, req_height) | ||
|
|
||
| -- Best fit (in most cases, we're operating on a best-fit scale factor already, so both sides should match). |
There was a problem hiding this comment.
Still asking :) "in most cases" - did you meet cases where both sides did not match ?
Also, I dunno about the word "best-fit" : we always keep aspect ratio, we just scale, always keeping the original aspect ration - and we start the image scaled so it's fully contained in the screen, A/R preserved.
There was a problem hiding this comment.
No, but I haven't checked many IV callers ;).
(Best-fit is the wording used in the original code; because there is a stretch codepath, but, IIRC, it's only used by the Screensaver and the Screensaver-like plugins).
There was a problem hiding this comment.
And those callers should never trip manual zooms ;).
There was a problem hiding this comment.
Oh, ok, you thought about these then. Logic splitted between simple ImageViewer and twisted serves-all-use-cases ImageWidget doesn't make things clear :/
|
I trust you on the user experience with pinch/spread. |
Those are currently always computed relative to the screen's dimensions (which sorta makes sense given the gesture), I'll see if I can fudge it similarly (because otherwise we're back to needing seven gestures to zoom in on low factors ;p). |
Also, update the zone to match GestureManager's *_edge zones. (Which matches how wide my tiny fingers actually are ;p).
As it apparently doesn't make the intent clearer ;).
|
OK, tested on my Kobo: feels a lot less frustrating, and feels rather alright. Probably unrelated to these changes, and I can't reproduce it on the emulator: zooming in on my Kobo on the images at page 20 (some biology stuff in French) and page 58 (map of the world with flags) crashes (before your max cap stops me) with something I have never seen: |
Yup, same general feeling here, FWIW. When the distance is relative to the screen, it feels better when zooming in (only a tad slow), but it's extremely irritating when zooming out from small factors. (What I mean to say is that this is a compromise that feels okay in most cases, compared to stuff that might feel better in some cases but very much worse in others).
We can always bump up the heuristics later, I ended by testing on an Elipsa, where we have RAM to spare, and it's definitely on the low end of things, so we should have a bit of leeway ;).
I had one crash inside MuPDF during testing, too (on the H2O, at high scaling, so it might have been caused by memory exhaustion somewhere inside mupdf during scaling. Or not ;p). |
|
Alas, not much help from your gdb on my kobo:
I guess it's fine. when |
You have to set a break on throw or something to catch a C++ exception, IIRC. |
Ok, this worked: So, probably similar to a malloc() failing: |
|
Okay, so maaaaaybe not so much breathing room after all :D. |
There was a problem hiding this comment.
Reviewed 1 of 5 files at r1, 2 of 4 files at r4, 1 of 2 files at r11, 1 of 1 files at r12, all commit messages.
Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @poire-z and @zwim)
frontend/cache.lua line 170 at r3 (raw file):
Previously, zwim wrote…
Additionally, I don't know what would happen if you call
memfree /memtotal(when they come wrong lshifted fromcalcFreeMem).
Done.
There was a problem hiding this comment.
Reviewed 1 of 1 files at r13, all commit messages.
Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @poire-z and @zwim)
|
I'm playing with another approach: zooming to the exact dimension that matches the distance between the final position of the two fingers. That possibly feels more natural, or at least more like what would happen if the rescaling was live and followed your fingers, like a corner drag to resize a window. It's... mildly fun. Haven't played with it much, though. It requires tweaking how GestureDetector computes the gesture distance, which may be an issue for other pinch users (it currently uses the travel distance of both fingers). diff --git a/frontend/device/gesturedetector.lua b/frontend/device/gesturedetector.lua
index 80d460d62..361bd77a6 100644
--- a/frontend/device/gesturedetector.lua
+++ b/frontend/device/gesturedetector.lua
@@ -1078,6 +1078,7 @@ function Contact:handleTwoFingerPan(buddy_contact)
ges_ev.ges = "outward_pan"
end
ges_ev.direction = gesture_detector.DIRECTION_TABLE[tpan_dir]
+ ges_ev.distance = end_distance
elseif self.state == Contact.holdState then
ges_ev.ges = "two_finger_hold_pan"
-- Flag 'em for holdState to discriminate with two_finger_hold_release
diff --git a/frontend/ui/widget/imageviewer.lua b/frontend/ui/widget/imageviewer.lua
index b6e62e084..1ddaf9c3b 100644
--- a/frontend/ui/widget/imageviewer.lua
+++ b/frontend/ui/widget/imageviewer.lua
@@ -637,6 +637,7 @@ end
-- Zoom events
function ImageViewer:onZoomIn(inc)
+ logger.dbg("ImageViewer:onZoomIn", inc)
if self.scale_factor == 0 then
-- Get the scale_factor made out for best fit
self.scale_factor = self._scale_factor_0 or self._image_wg:getScaleFactor()
@@ -671,6 +672,7 @@ function ImageViewer:onZoomIn(inc)
end
function ImageViewer:onZoomOut(dec)
+ logger.dbg("ImageViewer:onZoomOut", dec)
if self.scale_factor == 0 then
-- Get the scale_factor made out for best fit
self.scale_factor = self._scale_factor_0 or self._image_wg:getScaleFactor()
@@ -706,6 +708,97 @@ function ImageViewer:onZoomOut(dec)
return true
end
+function ImageViewer:onZoomToHeight(height)
+ logger.dbg("ImageViewer:onZoomToHeight", height)
+ if self.scale_factor == 0 then
+ -- Get the scale_factor made out for best fit
+ self.scale_factor = self._scale_factor_0 or self._image_wg:getScaleFactor()
+ end
+
+ local new_factor = height / self._image_wg._img_h
+
+ if not self._min_scale_factor or not self._max_scale_factor then
+ self._min_scale_factor, self._max_scale_factor = self._image_wg:getScaleFactorExtrema()
+ end
+ -- Clamp to sane values
+ new_factor = math.min(new_factor, self._max_scale_factor)
+ new_factor = math.max(new_factor, self._min_scale_factor)
+ if new_factor ~= self.scale_factor then
+ self.scale_factor = new_factor
+ self:update()
+ else
+ if self.scale_factor == self._min_scale_factor then
+ logger.dbg("ImageViewer:onZoomOut: Hit the min scaling factor:", self.scale_factor)
+ else
+ logger.dbg("ImageViewer:onZoomOut: No change in scaling factor:", self.scale_factor)
+ end
+ end
+ return true
+end
+
+function ImageViewer:onZoomToWidth(width)
+ logger.dbg("ImageViewer:onZoomToWidth", width)
+ if self.scale_factor == 0 then
+ -- Get the scale_factor made out for best fit
+ self.scale_factor = self._scale_factor_0 or self._image_wg:getScaleFactor()
+ end
+
+ local new_factor = width / self._image_wg._img_w
+
+ if not self._min_scale_factor or not self._max_scale_factor then
+ self._min_scale_factor, self._max_scale_factor = self._image_wg:getScaleFactorExtrema()
+ end
+ -- Clamp to sane values
+ new_factor = math.min(new_factor, self._max_scale_factor)
+ new_factor = math.max(new_factor, self._min_scale_factor)
+ if new_factor ~= self.scale_factor then
+ self.scale_factor = new_factor
+ self:update()
+ else
+ if self.scale_factor == self._min_scale_factor then
+ logger.dbg("ImageViewer:onZoomOut: Hit the min scaling factor:", self.scale_factor)
+ else
+ logger.dbg("ImageViewer:onZoomOut: No change in scaling factor:", self.scale_factor)
+ end
+ end
+ return true
+end
+
+function ImageViewer:onZoomToDiagonal(d)
+ logger.dbg("ImageViewer:onZoomToDiagonal", d)
+ if self.scale_factor == 0 then
+ -- Get the scale_factor made out for best fit
+ self.scale_factor = self._scale_factor_0 or self._image_wg:getScaleFactor()
+ end
+ -- Enter trig! c.f., https://math.stackexchange.com/a/3369637
+ local r = self._image_wg:getCurrentWidth() / self._image_wg:getCurrentHeight()
+ local h = math.sqrt(math.pow(d, 2) / (math.pow(r, 2) + 1))
+ local w = h * r
+ logger.dbg("Current factor:", self.scale_factor)
+ logger.dbg("Current: w =", self._image_wg:getCurrentWidth(), "h =", self._image_wg:getCurrentHeight(), "d =", self._image_wg:getCurrentDiagonal(), "r =", self._image_wg:getCurrentWidth() / self._image_wg:getCurrentHeight())
+ logger.dbg("New: w =", w, "h =", h, "d =", d, "r =", w / h)
+
+ local new_factor = math.min(w / self._image_wg._img_w, h / self._image_wg._img_h)
+
+ if not self._min_scale_factor or not self._max_scale_factor then
+ self._min_scale_factor, self._max_scale_factor = self._image_wg:getScaleFactorExtrema()
+ end
+ -- Clamp to sane values
+ new_factor = math.min(new_factor, self._max_scale_factor)
+ new_factor = math.max(new_factor, self._min_scale_factor)
+ if new_factor ~= self.scale_factor then
+ self.scale_factor = new_factor
+ self:update()
+ else
+ if self.scale_factor == self._min_scale_factor then
+ logger.dbg("ImageViewer:onZoomOut: Hit the min scaling factor:", self.scale_factor)
+ else
+ logger.dbg("ImageViewer:onZoomOut: No change in scaling factor:", self.scale_factor)
+ end
+ end
+ return true
+end
+
function ImageViewer:onSpread(_, ges)
-- We get the position where spread was done
-- First, get center ratio we would have had if we did a pan to there,
@@ -719,36 +812,57 @@ function ImageViewer:onSpread(_, ges)
-- meaning using the image dimensions here takes less zoom steps to get it back to a sensible size;
-- *and* large scale factors (where the image dimensions are larger than the screen),
-- meaning using the screen dimensions here makes zoom steps, again, slightly more potent.
- local inc
if ges.direction == "vertical" then
- inc = ges.distance / math.min(Screen:getHeight(), self._image_wg:getCurrentHeight())
+ if ges.distance > self._image_wg:getCurrentHeight() then
+ self:onZoomToHeight(ges.distance)
+ else
+ self:onZoomIn(ges.distance / math.min(Screen:getHeight(), self._image_wg:getCurrentHeight()))
+ end
elseif ges.direction == "horizontal" then
- inc = ges.distance / math.min(Screen:getWidth(), self._image_wg:getCurrentWidth())
+ if ges.distance > self._image_wg:getCurrentWidth() then
+ self:onZoomToWidth(ges.distance)
+ else
+ self:onZoomIn(ges.distance / math.min(Screen:getWidth(), self._image_wg:getCurrentWidth()))
+ end
else
- local tl = Geom:new{ x = 0, y = 0 }
- local br = Geom:new{ x = Screen:getWidth() - 1, y = Screen:getHeight() - 1}
- local screen_diag = tl:distance(br)
- inc = ges.distance / math.min(screen_diag, self._image_wg:getCurrentDiagonal())
+ if ges.distance > self._image_wg:getCurrentDiagonal() then
+ self:onZoomToDiagonal(ges.distance)
+ else
+ local tl = Geom:new{ x = 0, y = 0 }
+ local br = Geom:new{ x = Screen:getWidth() - 1, y = Screen:getHeight() - 1}
+ local screen_diag = tl:distance(br)
+ self:onZoomIn(ges.distance / math.min(screen_diag, self._image_wg:getCurrentDiagonal()))
+ end
end
- self:onZoomIn(inc)
return true
end
function ImageViewer:onPinch(_, ges)
-- With Pinch, unlike Spread, it feels more natural if we keep the same center point.
-- Set some zoom decrease value from pinch distance
- local dec
if ges.direction == "vertical" then
- dec = ges.distance / math.min(Screen:getHeight(), self._image_wg:getCurrentHeight())
+ -- FIXME: Only if image is smaller than the screen?
+ if ges.distance < self._image_wg:getCurrentHeight() then
+ self:onZoomToHeight(ges.distance)
+ else
+ self:onZoomOut(ges.distance / math.min(Screen:getHeight(), self._image_wg:getCurrentHeight()))
+ end
elseif ges.direction == "horizontal" then
- dec = ges.distance / math.min(Screen:getWidth(), self._image_wg:getCurrentWidth())
+ if ges.distance < self._image_wg:getCurrentWidth() then
+ self:onZoomToWidth(ges.distance)
+ else
+ self:onZoomOut(ges.distance / math.min(Screen:getWidth(), self._image_wg:getCurrentWidth()))
+ end
else
- local tl = Geom:new{ x = 0, y = 0 }
- local br = Geom:new{ x = Screen:getWidth() - 1, y = Screen:getHeight() - 1}
- local screen_diag = tl:distance(br)
- dec = ges.distance / math.min(screen_diag, self._image_wg:getCurrentDiagonal())
+ if ges.distance < self._image_wg:getCurrentDiagonal() then
+ self:onZoomToDiagonal(ges.distance)
+ else
+ local tl = Geom:new{ x = 0, y = 0 }
+ local br = Geom:new{ x = Screen:getWidth() - 1, y = Screen:getHeight() - 1}
+ local screen_diag = tl:distance(br)
+ self:onZoomOut(ges.distance / math.min(screen_diag, self._image_wg:getCurrentDiagonal()))
+ end
end
- self:onZoomOut(dec)
return true
end |
That's what I had it mind it should be, so, good luck ! :)
|
|
FWIW, currently, distance is the sum of the travel of both fingers. I'm using the distance between both fingers at the end of the gesture instead. If you start or end from pinched (i.e., "kissing" ;p) fingers, the two roughly matches, but that's not always what the gesture looks like (as you can start/end with both fingers still apart, possibly quite far) ;). (Fun fact: noticed that this also applies to two_finger_swipe, where it makes absolutely no sense (i.e., for the same actual distance, a two finger swipe will report roughly twice the distance of the same swipe with a single finger :s); but existing code already had to correct for it, so I just documented the quirk). |
* ImageViewer: Minor code cleanups * GestureDetector: Fix the `distance` field of `two_finger_pan` & `two_finger_swipe` gestures so that it's no longer the double of the actual distance traveled. Get rid of existing workarounds throughout the codebase that had to deal with this quirk.
Regression since koreader#9529 Fix koreader#9565
KOReader 2022.10 "Muhara"  We skipped last month's release because I was right in the middle of moving, which serendipitously coincided with fairly drastic changes that needed more time for testing, such as a big rewrite of gestures and multitouch (koreader#9463). Users of the Dropbox plugin will now be able to use the new short-lived tokens (koreader#9496). <img width="40%" alt="image" src="https://user-images.githubusercontent.com/59040746/193070490-a3d477db-bd82-431b-95fd-2c4765244378.png" align="right">One of the more visible additions is the new Chinese keyboard contributed by @weijiuqiao, based on the [stroke input method](https://en.wikipedia.org/wiki/Stroke_count_method) (koreader#9572). It's not smart and it requires knowledge of stroke order. A tutorial can be found [here](https://github.com/koreader/koreader/wiki/Chinese-keyboard), part of which I will reproduce below. <hr> The stroke input method groups character strokes into five categories. Then any character is typed by its stroke order. | Key | Stroke type | | ------ | ------ | | `一` | Horizontal or rising stroke | | `丨` | Vertical or vertical with hook | | `丿` | Falling left | | `丶` | Dot or falling right | | `𠃋` | Turning | For example, to input 大, keys `一丿丶` are used. Note all turning strokes are input with a single `𠃋` key as long as they are written in one go. So 马 is input with `𠃋𠃋一`. After getting the intended character, a `分隔`(Separate) or `空格`(Space) key should be used to finish the input. Otherwise, strokes of the next character will be appended to that of the current one thus changing the character. Besides, the keyboard layout contains a wildcard key `*` to use in place of any uncertain stroke. Swipe north on the `分隔`(Separate) key for quick deletion of unfinished strokes. <hr> Logo credit: @bubapet We'd like to thank all contributors for their efforts. Some highlights since the previous release include: * NewsDownloader: Strip byte order mark from xml string before parsing (koreader#9468) @ad1217 * GestureDetector: Full refactor for almost-sane(TM) MT gesture handling (koreader#9463) @NiLuJe * Kobo: Unbreak touch input on fresh setups on Trilogy (koreader#9473) @NiLuJe * Kobo: Fix input on Mk. 3 (i.e., Kobo Touch A/B). (koreader#9474, koreader#9481) @NiLuJe * Kindle: Attempt to deal with sticky "waking up" hibernation banners (koreader#9491) @NiLuJe * Add "Invert page turn buttons" to Dispatcher (koreader#9494) @NiLuJe * [UIManager] Outsource device specific event handlers (koreader#9448) @zwim * AutoWarmth: add a choice to control warmth and/or night mode (koreader#9504) @zwim * Allow F5 key to reload document (koreader#9510) @poire-z * bump crengine: better SVG support with extended LunaSVG (koreader#9510) @poire-z * CRE/ImageViewer: get scaled blitbuffer when long-press on SVG (koreader#9510) @poire-z * RenderImage: use crengine to render SVG image data (koreader#9510) @poire-z * Wikipedia EPUBs: keep math SVG images (koreader#9510) @poire-z * TextViewer: add Find (koreader#9507) @hius07 * A random assortment of fixes (koreader#9513) @NiLuJe * Add Russian Wiktionary dictionary (koreader#9517) @Vuizur * add custom mapping for tolino buttons (koreader#9509) @hasezoey * Profiles: add QuickMenu (koreader#9526) @hius07 * ImageViewer: Clamp zoom factor to sane values (koreader#9529, koreader#9544) @NiLuJe * ReaderDict: fix use of dicts with ifo with DOS line endings (koreader#9536) @poire-z * Kobo: Initial Clara 2E support (koreader#9545) @NiLuJe * TextViewer: add navigation buttons (koreader#9539) @hius07 * ConfigDialog: show button with default values in spinwidgets (koreader#9558) @hius07 * Misc: Get rid of the legacy defaults.lua globals (koreader#9546) @NiLuJe * Misc: Use the ^ operator instead of math.pow (koreader#9550) @NiLuJe * DocCache: Unbreak on !Linux platforms (koreader#9566) @NiLuJe * Kobo: Clara 2E fixes (koreader#9559) @NiLuJe * Keyboard: add Chinese stroke-based layout (koreader#9572, koreader#9582) @weijiuqiao * Vocabulary builder: add Undo study status (koreader#9528, koreader#9582) @weijiuqiao * Assorted bag'o tweaks & fixes (koreader#9569) @NiLuJe * ReaderFont: add "Font-family fonts" submenu (koreader#9583) @poire-z * FileManager: add Select button to the file long-press menu (koreader#9571) @hius07 * Dispatcher: Fixes, Sort & QuickMenu (koreader#9531) @yparitcher * Cloud storage: add Dropbox short-lived tokens (koreader#9496) @hius07 * GH: Extend the issue template to request verbose debug logs for non-crash issues. (koreader#9585) @NiLuJe * Logger: Use serpent instead of dump (koreader#9588) @NiLuJe * LuaDefaults: Look for defaults.lua in $PWD first (koreader#9596) @NiLuJe * UIManager: Don't lose track of the original rotation on reboot/poweroff (koreader#9606) @NiLuJe * ReaderStatus: save status summary immediately on change (koreader#9619) @hius07 * [feat] Add Thai keyboard (koreader#9620) @weijiuqiao * Dispatcher: Fix subtle bug with modified items being added twice to the sort index (koreader#9628) @yparitcher * Vocabulary builder: supports review in reverse order (koreader#9605) @weijiuqiao * Exporter plugin: allow adding book md5 checksum when exporting highlights (koreader#9610) @sp4ke * buttondialogtitle: align upper borders (koreader#9631) @hius07 * Kobo: Always use open/write/close for sysfs writes (koreader#9635) @NiLuJe * OPDS-PS: Fix hardcoded namespace in count (koreader#9650) @bigdale123 [Full changelog](koreader/koreader@v2022.08...v2022.10) — [closed milestone issues](https://github.com/koreader/koreader/milestone/59?closed=1) --- Installation instructions: [Android](https://github.com/koreader/koreader/wiki/Installation-on-Android-devices) • [Cervantes](https://github.com/koreader/koreader/wiki/Installation-on-BQ-devices) • [ChromeOS](https://github.com/koreader/koreader/wiki/Installation-on-Chromebook-devices) • [Kindle](https://github.com/koreader/koreader/wiki/Installation-on-Kindle-devices) • [Kobo](https://github.com/koreader/koreader/wiki/Installation-on-Kobo-devices) • [PocketBook](https://github.com/koreader/koreader/wiki/Installation-on-PocketBook-devices) • [ReMarkable](https://github.com/koreader/koreader/wiki/Installation-on-ReMarkable) • [Desktop Linux](https://github.com/koreader/koreader/wiki/Installation-on-desktop-linux) • [MacOS](https://github.com/koreader/koreader/wiki/Installation-on-MacOS)
Computed with magic ;p
(Seriously, don't ask me to explain the maths :D).
This change is