M92: 16 level gray-scales image distortion

[Source]

Even though displaying of images is not the primary purpose of M92 it still does a nice job showing gray-scale images. If there wouldn’t be a little distortion of the shade smoothness it could be just perfect.

Viewing images a lot, I have noticed that the darkening steps of the 16 gray tones are not smoothly distributed between black and white. With other words, if you display all 16 shades of gray on a palette there is a smooth increase of darkness for the first 11 levels (from white to black), then there is a sudden disproportionately increased darkening between 11th and 12th. Again follows smooth darkening between 12,13,14th levels, and a second unwanted sudden increased darkening between 14 and 15 levels. The transition from 15 to 16 is smooth again.

You can see what I mean if you take a look at the attached image gray_scales_on_M92.jpg. This is the photo taken of M92 displaying the image gray_scales.gif. Basically there are two disproportionately large jumps of darkening, while there are some adjacent light shades hardly distinguishable. Unfortunately this results in a strange photo viewing experience. If the photo is light it will produce a very nice picture where the shades from 1 to 11 are used. Then if there is a slightly darker area there will be a sudden jump of darkness sometimes creating a caricature-like distortion.

This can be seen on the sample_on_M92.jpg photo where the full_color_sample.jpg is displayed on M92. The left corner and edge of the mouth is exaggeratedly darkened. The same effect is visible at the corner of the eye on the left and the edge of the face and hair. It creates an impression of badly developed black&white photo, or a distorted image. If you convert the original full_color_sample.jpg into 4bit gray-scale (16_level_gray_sample.gif), and display it on a PC screen these distortions will not be visible. It is not too bad on this photo, but it is quite disturbing on darker photos. This is definitely a bug in M92. Most probably a software bug that uses an uneven gray-scale level distribution.

Did anybody else notice this phenomenon yet?

If it really turns out to be a software bug, I would really appreciate if the developer community or Onyx could fix it because this would make the difference between fairly usable vs. perfect photo quality image display. I am sure there are many users like myself who use the M92 for viewing photos, graphs, and drawings, so it is not an unimportant feature. In my case in fact, the ability to display 16 gray levels and a large screen were important factors while deciding to buy this E reader.

Cheers,
Joe

[Mono]

I read on Russian forum that there is a callibration curve translating the picture shades to pixel shades.

It was possible to load to device a new curve. I am not sure if it was M92 or other Onyx product.

Generaly. It seems it is possible somehow to change the translation table (callibration curve). It is hard to say, if the behaviour, you discovered is intentional or it is sort of bug in table. In the case discussed at forum, there was by mistake in some readers wrong callibration curve and support helped the user to change it to right one.

Only Onyx could tell us, if this is a desired feature (making reading text easier) or some kind of mistake.

It is not an real issue for me, if I haven't noticed it untill now.

EDIT.
It looks like the output has only 14 shades, not 16 shades of display...

[JSWolf]

The dithering routine(s) seem to be not well written.

[Source]

Originally Posted by Mono

It looks like the output has only 14 shades, not 16 shades of display...

Yes, You are right. Now that I took a closer look, indeed there is no difference between the 10&11th shades and between 12&13th shades. So basically this is definitely a bug either in software or the display itself. It can't display 16 shades but only 14.

Can you please, (or anybody else) try to display the gray_scales.gif (attached in first post) on your M92 and check if you have the same problem? Or maybe only my device is faulty and needs to be returned...

Thanks.

[janek]

Originally Posted by Source

Can you please, (or anybody else) try to display the gray_scales.gif (attached in first post) on your M92 and check if you have the same problem? Or maybe only my device is faulty and needs to be returned...

Thanks.

I confirm that. 10/11th and 12/13th shade look the same.

[Mono]

Originally Posted by Source

Can you please, (or anybody else) try to display the gray_scales.gif (attached in first post) on your M92 and check if you have the same problem? Or maybe only my device is faulty and needs to be returned...

Strange, my M92 does show 16 grades of gray. I may distinguish every shade box and none is the same like another.

That might explain, I haven't noticed anything strange concernimg b&w pictures.

My M92 is probably from 1st batch for European market. It arrived in December after Christmas. My current firmware is the lates one - 1.7 - 20120726.

In your case the problem may be hardware or "software" the callibration curve. Hope Booxtor might help to all, if it is the "software" one.

I do not remember exactly, what had to be done in case discussed on a Russian forum, but my feeling is it was not only simple copying the file somewhere to a normal firmware. My feeling is, the callibration table must be copied to a deeper "firmware" layer than normal firmware is written. Maybe somewhere to the screen hardware.

I do not know, just guessing from the fact, that the solution recomended was not to inslall latest (or any other known correct) firmware, instead certain utility (provided by device support) had to be run.

[Source]

Thanks people for checking out this issue.

I have just now bought my M92 from Booxtor, and have good reason to think that it is from the latest batch. I have also upgraded the firmware to the latest 1.7.20120726, though had the "boot freezing" problem after restart. So I had to install the special firmware update posted at https://www.mobileread.com/forums/sho...8&postcount=36 to get it functional again. Then reinstalled the latest normal update on top of that again.

It would be nice if Booxtor could check whether this fault is present in all devices in the latest delivery batch, or only some E readers (like mine and Janek's) have it as a result of poor quality control at factory.

Mono, It would be great if you could find the relevant posts (or at least the thread) on the Russian forum you mentioned. I don't speak Russian, but if only few posts need to be translated I can do that with Google translator.

It is important for me to get this bug fixed, because it really spoils the darker photos and graphics.

[Mono]

I will try to find.

I was able to speak/read Russian in very past, but now I use Google translator. After more than 25 years of not using the language at all my knowledge is almost gone....

EDIT:
I cannot find it.

Try to ask Dmitry Shkarin the developer of OTFM file manager. He might know, or he might ask in Russian forum.

At Russian forum:
http://translate.google.com/translat...1f47672b1618d8

Or this forum:

https://www.mobileread.com/forums/sho...d.php?t=184789

[Booxtor]

Hi,
I can confirm this picture in the last batch of M92.
Current waveform is V220_C023_97_WJ7901_P1_BTC
The new improved waveform is V220_C024_97_WJ9601_P1_BTC

there are some tricky ways to get the new waveform to the M92, but I beg you to wait a little bit - update is comming soon

[onyxuser]

booxtor, so it's a firmware issue, not hardware?

If I buy device with this issue, can I fix it with a firmware update?

Can you explain what means "waveform is V220_C023_97_WJ7901_P1_BTC" or "waveform is V220_C024_97_WJ9601_P1_BTC" ?

[Mono]

Originally Posted by Booxtor

there are some tricky ways to get the new waveform to the M92, but I beg you to wait a little bit - update is comming soon

OK, if I understand it well. It is not a hardware issue, but sort of soft one which can be solved. It seems to me similar case to one on a Russian forum (they also refered to labels with many numbers and letters...).

EDIT.
This is not the exact thread I read on Russian forum months ago and I have reffered to in previous posts, but it is also relevant.
http://www.the-ebook.org/forum/viewt...7d6e638c1014be

[Booxtor]

Originally Posted by Mono

OK, if I understand it well. It is not a hardware issue, but sort of soft one which can be solved. It seems to me similar case to one on a Russian forum (they also refered to labels with many numbers and letters...).

EDIT.
This is not the exact thread I read on Russian forum months ago and I have reffered to in previous posts, but it is also relevant.
http://www.the-ebook.org/forum/viewt...7d6e638c1014be

Yes, it can be changed by software.

If someone can't wait look at this post: http://translate.google.com/translat...925302&act=url
(click "+" to expand description!)

To be honest I would never notice that issue, if I was not explicit pointed to it. Thanks for it!
AFAIK in text documents e-ink devices switch to 4 grade grey anyway. 16 grade grey is only available in picture viewers.

[Mono]

Originally Posted by Booxtor

AFAIK in text documents e-ink devices switch to 4 grade grey anyway. 16 grade grey is only available in picture viewers.

Booxtor, is that true (4 grade grey) also for scanned book pdfs and pictures/graphs in normal text pdfs, or just the text?

[guidol]

Originally Posted by Source

If you convert the original full_color_sample.jpg into 4bit gray-scale (16_level_gray_sample.gif), and display it on a PC screen these distortions will not be visible.

For converting Pictures to a e-ink Reader there is a nice page about the topic:
http://www.ebookworm.us/2011/05/31/d...-saver-images/

And here is a 16-Gray-PlugIn for GIMP:
http://www.ebookworm.us/wp-content/u...aysSixteen.zip

[Source]

Originally Posted by Booxtor

Hi,
I can confirm this picture in the last batch of M92.
Current waveform is V220_C023_97_WJ7901_P1_BTC
The new improved waveform is V220_C024_97_WJ9601_P1_BTC

there are some tricky ways to get the new waveform to the M92, but I beg you to wait a little bit - update is comming soon

Thanks Booxtor for the info and advice.

I would be most happy to skip the nontrivial fixing with Mfgtools, and wait for the next update if it would be released within a couple of weeks. But if it is uncertain, or it takes many weeks and months, then I will try to figure out how to apply the patch. I want to use it now, and this bug spoils the image viewing usability.

When can we expect the new update including the gray-scale fix?

I have checked my Screen Waveform is: V220_C073_97_WNA301_ED097OC4_BTC which is different than what you mentioned as "current Waveform".