More display issues on the DR

[Grimulkan]

In the spirit of previous posts, here are a few test images comparing the Sony PRS-505 and the DR1000S displays.

Before I begin, let me say that I love my DR and wouldn't dream of replacing it with anything in the market right now.

That said, I am generally disappointed in the DR1S, especially for tone reproduction. This would be of concern to anyone who uses the DR for viewing photography journals or perhaps even reading graphic novels/manga, but should not affect reading of plain text. The good news is, I think the issues can be corrected via software updates, either from iRex or from the community. Perhaps someone (maybe me?) porting/writing a CBR viewer can look into this.

In the screenshots below, look for differences between the 2 devices, rather accept these as an absolute reproduction, as the camera and your monitor influence the images. The dpi on both devices is about the same (the Sony has a lower resolution but a smaller screen), but the Sony has only 8 shades of gray as opposed to the DR's 16.

I've included a zip of all the original test images used (download original_test_images.zip) as well as a copy of the screenshots (download screenshots.zip).

1. First off, the Sony display does seem to have a slightly "whiter" whites.



This may seem like nitpicking, but it actually produces a noticeable difference in practice, even when reading plain text.

2. Here is a screenshot of a 16-shade gray picture (link to gray16.gif) on both devices:



The Sony is simply fantastic, especially given that it has only 8 shades of gray! I think this is some brilliant dithering on the part of Sony, you can actually count 16 seemingly distinct bands. The DR, on the other hand, uses some strange algorithm and actually runs some of the bands together.

Here is the same image when "embedded" onto an html file and viewed on FBReader:



There are actually 16 distinct shades on the DR this time (as there should be), and looks close to what the Sony produces. The good news is that this means that the DR runs a software algorithm (in the UDS), and can possibly be corrected in updates/third-party software.

3. It gets worse. This is a grayscale gradient with 256 shades (link to grad.jpg).



The Sony simply blows me away, that is some amazing dithering! You see no banding despite the lower number of shades. On the DR, you not only see distinct bands of colors, but also have dark/light bands where there shouldn't be (the image *should* go from lightest to darkest).

Here is the same thing on FBReader:



FBReader does not do any dithering, and you can see 16 bars. This is not as good as the Sony, but at least there are no strange brightness/darkness interchanges.

4. Here is a photograph (link to baby.jpg) with a lot of subtle tonal gradations:



Look at the baby's cheeks, the 16 shades of the DR is just wasted with poor dithering
FBReader image reproduction (from 2 and 3) is more "faithful", but does no intelligent dithering, and does not match up to the Sony, or even the UDS.



This means that intelligent processing can do a world of good, but what the DR does (with UDS) is not as good as the Sony (2, 3, and 4) and can sometimes be plain wrong (see 3). Given that the Sony has an inferior processor compared to the DR (as well as a superior battery life), I don't think running a smarter algorithm on the DR should be any more demanding on its current hardware, if at all.

5. All is not lost, here is a high contrast image which looks about the same on both readers:



Images like this (or plain text) should come out looking fine, but you're really not exploiting those wonderful extra shades of gray here.

6. Other thoughts:
There seems to be significantly more "ghosting" on the DR (you can see traces of the dreaded "please wait" window, most apparent in the UDS in 3). I don't know if this is a hardware issue or merely due to the larger screen size. Either way, this doesn't bother me too much, but I would enjoy a "ghostbuster" button that simply refreshed everything and produced a clean image when I wanted one

Finally, here is the test setup:



I would like to thank my own capacity for unlimited procrastination as well as the joblessness of my roommate, without which none of this would have been possible

Update 05/19: I've thought about the Sony being "whiter" long and hard, and I can't put my finger on it. The Sony display has more "pop" to it, perhaps due to whiter whites, deeper blacks, or the silver casing causing the eye to perceive things this way. This is also seen in the last test image on this post (the one with the lamppost), where the DR looks "warmer" and more "yellow", while the Sony is "cooler" and more "blue". You can also see the difference in the leaves on the tree. This doesn't seem to have anything to do with dithering.

I have tried this out with two different Sony readers, and my impression stands. It is still possible that there is something sub-par about my specific DR (such as my Vcom being off), so I don't know. I am going to try enclosing my DR in off-white paper and see if there is a perceived difference.

Update: For impressions regarding this color difference, ghosting, refresh rates and manga reading, look further in this thread

[RobbieClarken]

That is some really impressive work! I hadn't even noticed there was a problem but those pictures make a pretty convincing case. Hopefully some clever programmers will be able to improve the rendering algorithm because I'm so jealous of Sony's dithering now.

[Grimulkan]

I would really appreciate if someone could post screenshots of the 3 test images (the 2 grayscale images: grad.jpg and gray16.gif as well as the baby picture) on different e-readers. I am most interested in what the Iliad does; if it looks better then it means that iRex actually already knows how to fix this issue, and perhaps will in a future update.

[Matthijs]

This is purely software (dithering algorithm) and can be fixed in a SW upgrade. This has our attention, but I cannot promise a time frame.

If anyone feels like looking at it, the image plugin for UDS is completely open source

[HarryT]

Until this does get fixed, you're probably best off doing the conversion to grey-scale in a decent graphics package before transferring the file to the DR. I realise that this is of no help when the image is a part of a PDF or other file, however.

[Grimulkan]

Originally Posted by HarryT

Until this does get fixed, you're probably best off doing the conversion to grey-scale in a decent graphics package before transferring the file to the DR. I realise that this is of no help when the image is a part of a PDF or other file, however.

Actually, you can batch-process image-only pdfs as well (using Imagemagick for instance).

[allovertheglobe]

These were originally part of my "Sketching Goodies" thread a while back. I pulled them when I refocused on the actual writepads. Load them on your DR1000 and examine them for yourself. The original has a smooth 256 gray gradient. The 16 is already pre-reduced to see how the grays are remapped. The lower strip is good for checking for moire patterns or scaling issues (PDFs anyone...) I used these useful to diagnose the weird "light/wrong grayscale" issue I have had a couple of times.

I was going to look into remapping the 16 grays in such a way that an even scale would result (in Photoshop or similar) but if it can be fixed in firmware for all images in general, all the better.

I have to say though: in everyday use (manga reading ) it is not nearly as noticeable as in these specific examples.

P.S.: Two big thumbs up for any CBR / compressed image folder viewing app!!

[nrapallo]

Originally Posted by Grimulkan

I would really appreciate if someone could post screenshots of the 3 test images (the 2 grayscale images: grad.jpg and gray16.gif as well as the baby picture) on different e-readers.

Using PDFRead and the 4 original_test_images, here are some ebooks I created to load onto different readers.

Upload your screenshots to compare with others...

[cbell]

I'm watching this thread with a lot of interest.

I can view the files on several devices, but I just don't have the photography skills to do them justice.

[pwjone1]

I wonder if the differences have to do with the touch/stylus layer on top of the normal e-ink screen for the DR1000S. The reason I ask, is having looked at the Sony PRS-505 and PRS-700 side-by-side, the PRS-505 is superior by far in terms of screen quality (contrast, viewing angle, etc.).

[Gertjan]

The touch/stylus layer (wacom sensor) is behind the screen so it does not affect the display. Actually, the screen quality of the DR1000 is generally considered top notch and the best e-ink display currently available. The grayscale issues in this thread are, as Matthijs mentioned, related to the dithering in software which indeed leave room for improvement.

[pwjone1]

Originally Posted by Gertjan

The touch/stylus layer (wacom sensor) is behind the screen so it does not affect the display. Actually, the screen quality of the DR1000 is generally considered top notch and the best e-ink display currently available. The grayscale issues in this thread are, as Matthijs mentioned, related to the dithering in software which indeed leave room for improvement.

Thanks, then it's a mystery then. Looking at picture #1, first entry, the DR1000S sure looks "greyer" than the Sony PRS 505, and that's exactly the case also with the PRS 700. But if it is just the dithering, hard to explain the white whites on the Sony (sorry, sounded like a detergent commercial for a second). They should be in theory ruffly the same gen e-ink display technology, just different sizes.

[olis]

Looking at picture #1, first entry, the DR1000S sure looks "greyer" than the Sony PRS 505,

I have found this picture and it seems to me that there isn't any difference in contrast ratio (background whiteness versus print blackness) between the prs-505 and the dr1000s.

[nrapallo]

Originally Posted by olis

I have found this picture and it seems to me that there isn't any difference in contrast ratio (background whiteness versus print blackness) between the prs-505 and the dr1000s.

Could you try displaying any of the test images in the ebooks above using the 16 grayscale (already dithered) images and the 256 grayscale (device dithered) images as comparisons.

[olis]

Could you try displaying any of the test images in the ebooks above using the 16 grayscale (already dithered) images and the 256 grayscale (device dithered) images as comparisons.

I'm afraid, I don't have the devices showed in the photo I posted. I found that picture surfing the net.