01 July 2014

On Wiki Loves Earth Brasil 2014

I was invited to be a member in the jury for the Wiki Loves Earth 2014 photography contest in Brasil and I joined gladly: it was both a honor and an opportunity to see some places which otherwise I don't expect to see any time soon. Also it was an opportunity to share from my experience with 3 editions already of the similar Wiki Loves Monuments in Romania. Now, as the Brasilian contest has published its winners, I want to share a few conclusions.
First, it was a success from any point of view, from the impressive number of images (around 7000), to the exposure the team managed to receive, with presences on the most important technology and photography magazines and portals in the country. Then, a lot of impressive places I wish I could see in person, a good selection of winners and a first place image I really like and envy the photographer who took it.

Amanhecer no Hercules --
'Amanhecer no Hercules' by Carlos Perez Couto, CC-BY-SA, winner of WLE 2014 Brasil

Unfortunately, nothing is perfect and I have a couple of pet peeves about the contest. First, not once I found myself shouting in my head 'They didn't hear of the rule of thirds in Brasil?' I saw too many otherwise nice pictures with the horizon line cutting the image in half for no good reason, which could be way better with a 'correct' composition. Second, I had the feeling some participants misunderstood the purpose of Wikimedia Commons and sent unedited images: is OK to upload thereedited images as long as you do not change the reality. I encountered a lot of photos which can easily be improved with a few simple edits: framing (crop and/or rotation), color/exposure adjustments (curves, levels, brightness, saturation), sharpening and such. Some photos could made the top 10 if they had be edited as such. Fortunately, grace to the license, such improvements can be made at a later date by any contributor.
Still, the Wiki loves Earth 2014 contest didn't end: is still ongoing in a few other countries and will conclude with an international contest. Best of luck to all participants! Have good photos and all of us will win (including myself as an ordinary Wikipedia user).

30 April 2014

Pop Art

Recently I took a very colorful and quite abstract picture, which I thought would make for an interesting 'pop art' effect. The process is really basic and obvious, but I decided to share it for anyone who want to learn a quickie.

pop art gimp

So, I opened the image with GIMP. Since I want the final collage as a 4x4 composition, increase the Canvas Size to 200% on both directions.

pop art gimp

Then Duplicate the image layer.

pop art gimp

Repeat the duplication until there are enough pieces to cover the image. I need 3 duplicates, for a total of 4 pieces.

pop art gimp

Select each piece and with the Alignment Tool move them to cover the image (one right, one bottom, one right and bottom).

pop art gimp

Now the aligned pieces should fill the entire image.

pop art gimp

Leave one layer as is (if you really want, you can edit it too) and for the second open the Hue-Saturation dialog.

pop art gimp

Move the Hue slider left or right until you are happy with the new color set.

pop art gimp

Repeat for the other layers until you have something like this:

pop art gimp

Export and you are done:

pop art gimp

Here's a different use case for a similar effect: I had a single background for the water drop photos, but adjusting the Hue made it appear the pictures are more different than in reality.

pop art gimp

PS: as someone told me, I should print this at some big size and try to sell my 'pop art' creation for a ginormous amount of money.

Firefox 29

firefox 29
Probably everybody already heard there is a new Firefox release and it changes many things and already has an opinion about it. Unfortunately I don't have the audience to gather a significant sample, but if anyone runs such a poll, I would be genuinely interested.

Which Firefox user interface do you prefer?
  1. Firefox 29 (the latest, just released)
  2. Firefox 4-28 (still on long term support)
  3. Firefox pre-4 (before the "keyhole")
  4. Seamonkey (the suite)
  5. I don't care (as long at it open web pages)
Of course i am coming here from an angle: I prefer the apps on my desktop to be consistent with each other, as layout, structure, widgets placement, shortcuts, behavior( I was happy the old Firefox never got the "keyhole" on Linux.) Still, not enough to move to something like Midori , where the look and feel is totally native but the set of features smaller.

22 April 2014

Pseudo-HDR editing

Usually I don't edit much my landscape photos, not because I don't know how but I prefer them this way. Still, recently I felt the need for some more advanced processing for a picture, it enjoyed some success so I decided to share the process. The tools used were UFRaw (in the form of the GIMP plugin), Luminance HDR and, of course, GIMP.

I passed by this scene in the nearby park at the "golden hour" and it looked photogenic, but I wanted to make it more dramatic. One can increase the drama in a landscape photo by using a HDR treatment, but not having the tripod with me (for a proper HDR image you need at least 3 images with exactly the same scene but different exposures) I decided to go for pseudo-HDR. For this, I set the camera recording mode to RAW.

pseudo hdr

Note: the real purpose of a HDR image is to have details both in the shadows and in the highlights, beyond what the camera sensor can record, the improved drama is a side effect.

The RAW image was imported in GIMP via the UFRaw plugin 3 times: with normal, -1 and +1 exposure. If you really want, you can try doing the same starting from a single JPEG an simulate the exposure bracketing with color levels/curves, but I wouldn't advise: if from a RAW you can recover some lost image details, in JPEG they are gone forever.

pseudo hdr

The result is 3 JPEG images, one under-exposed, one exposed properly and the other over-exposed, which are to be combined in a HDR. For more drama, you can bracket with more than one step.

pseudo hdr

I imported the JPEGs to Luminance HDR and set their exposures manually to -1, 0 and +1 (or whatever values you used for RAW development). Then just press "Next" a few times, there is no need to adjust parameters, nor align the images (they were obtained from the same source).

pseudo hdr

Now we have a High Dynamic Range image, which can't be used or viewed as-is on a normal computer display, it has to be converted back to Low Dynamic Range, but optimized for what do we want from it (details in shadows and/or highlights, drama, whatever).

pseudo hdr

Time to pick one of the presets in the right column, one you think is the best for your case.

pseudo hdr

Then I adjusted the color levels a bit (if you prefer, the levels can be adjusted later with GIMP or any other image editing app).

pseudo hdr

Now the image can be exported as a JPEG benefiting from the HDR/pseudo-HDR treatment. You can leave it as-is if you like.

pseudo hdr

However, I opened it again with GIMP for more refinement: sharpening and color curves adjustment, to make the colors warmer. This is my end result.

pseudo hdr

10 April 2014

A license to print money

We learned (link in Romanian) the Romanian government is negotiating with Microsoft for a Windows XP support extension in the public administration. This is the free market at work and there may be solid arguments for doing so (yes, I would prefer my tax money to be spent on Free and Open Source Software instead of making business with a convicted monopolist, but realistically I don't expect this to happen in the foreseeable future).
My source of amazement is: once the patches and the delivery infrastructure are set for the first government (UK), then the costs for adding any subsequent government is approaching zero: computers in Romanian administration usually have Windows installed in English, so there is nothing to translate, the infrastructure to deliver patches is there and the network costs are the same or lower (presumably patches fro XP are smaller in size as patches for Windows 7).
Apparently ending support for XP makes a lot of business sense for Microsoft, they suddenly started receiving important sums of money for things they did only a week ago for free. Who would refuse that?

08 April 2014

On the Mozilla scandal

Like many of you, I too received the Mozilla email about "helping" with communications regarding the "activities of the past week", which is an euphemism for the Brendan Eich scandal. It comes with a FAQ you are expected to quote when talking about the issue.

Since all communications from Mozilla on the topic were opaque, either in blog posts with comments disabled or emails from generic addresses where is useless to reply, I finally decided to write something. Anyway, April 1-st, when I learned about the topic, wasn't a good day to write about serious issues. It may be not me my business what happens inside the Mozilla corporation, but if the Mozilla community feels a need to guide my public communications about it, then maybe it is.

So I am totally unhappy with Mozilla to caving-in to pressure from bullies. It creates a serious precedent and a slippery slope. What's the next step? Once a new CEO is announced, no matter who he is, the browser will be blocked by anti-gay websites? It looks like we are going towards a Balkanized web, where the "best viewed with [Internet Explorer|Netscape Navigator]" buttons are replaced by "blocked for browsers whose CEO have private opinion X".

Seriously, instead of personal political opinions of the Mozilla CEO I was more concerned about him wasting resources on the mobile OS unicorn while the browser looks more and more like a Chrome copy-cat.

18 March 2014

DPI and photography

There is a good understanding of DPI among hardware geeks (they may boast about how superior is tablet X due to a higher DPI display), still I am surprised to see how many people from the photography world do not understand this (sometime don't want to learn, on the "is technical stuff, I am an artist and not care about technical details") to the point it becomes ridiculous, so I will try to explain it with simple words, in case someone will pay attention.

DPI stands for "Dots Per Inch" and is a characteristic of a hardware device (for example a computer/tablet/phone display). It says how many pixels are in one inch (1 inch = 2.54 cm). Example: the computer I use to write this piece has a 38 cm wide display, which is 38 cm / 2.54 ~= 15 inches. Considering the horizontal resolution is 1600 pixels, then it has a resolution of 1600 pixels / 15 = 106.67 DPI. Of course, the higher the DPI value, the better looking the image will be on your display, as it will enable to to see finer details.

dpi screen
In digital photography the situation is different: your photo is a file and it can be put on a wide range of displays. The DPI value is not as important as the actual image resolution in pixels, it is actually metadata. Actually the correct term when talking about photos is PPI, standing for "Points Per Inch", but PPI and DPI are close enough, is not a huge problem if you interchange them. PPI is relevant when you print the image, is the density of color points to be printed by a inch. The math is similar and having a target print size and print image quality (PPI), it will help determine the needed pixel resolution. Example: you want to print a 15 x 10 cm image at good quality (300 PPI). On the horizontal 15 cm / 2.54 = 5.9 inch, then 5.9 inch * 300 PPI = 1770 pixels. On the vertical, 10 cm / 2.54 = 3.96 inch, then 3.95 inch * 300 PPI = 1182 pixels. In conclusion, you will need a 1770 x 1182 pixels image. 1800 x 1200 is close enough and easy to remember, so a 1800 x 1200 image will print at good quality at 15 x 10 cm.
dpi print
JPEG is the file format we use day to day to exchange photos and it has the ability to store a DPI value somewhere inside its metadata, but is only metadata: an indication at which size (in cm or inches) to print the file. But you can print the same file at any dimension or any DPI/PPI value (of course, with the respective image quality consequence). Changing this value won't modify in any way the look and work of your digital file.

Now, what is a good DPI value for your print? This depends on its intended use, of course :) A 300 DPI is considered good enough for a quality print, like those in the glossy magazines, where you look closely and expect to see fine details. When printing a poster which will be seen from a couple of meters, you can lower the DPI value at 100 and for a billboard to be seen from tens of meters, you can go way lower: it does not mater the printed points are huge when looking closely, nobody will do that.

Now another practical example to illustrate the ridiculous part and how to deal with it. For a recent photo exhibition (it is still on display), the requirements were "100x66 cm at 240 DPI". This is ridiculous: 100 cm / 2.54 = 39.4 inch, 39.4 inch * 240 DPI = 9456 pixels and 66 cm / 2.54 = 26 inch, 26 inch * 240 DPI = 6240 pixels, so to satisfy it you need a 9456 x 6240 photo, which means 9456 * 6240 = 59005440 - you need a 60 Megapixel camera to produce it. Nobody in the target group for that expo has access to such a camera. What to do?

Knowing the people, I can safely say most of them just ignored a requirement they don't understand, and even if they understand can't follow. Still, some tried their best and this is the right thing to do, consider other exhibitions have sane requirements you can, and then should, follow, like the one asking for 1400 x 933 at 96 DPI.

The most obvious thing to do is to resize your image to achieve the needed resolution in both pixels and DPI (GIMP example below). This is sensible thing to do when you scale down the image, as in the 15 x 10 cm case, (reduce the pixel resolution count) and you can optimize interpolation method and post process your image. However, when it would need to scale up, as in the 100 x 66 cm case, is not only a waste of resources and time, extrapolation will lower the image quality so the result will be worse than printed at a low DPI value.
dpi print
What I do in such cases (and I got my images accepted in quite a few exhibitions) is to give the image at the largest pixel resolution I have available and then set the metadata DPI value for the desired target in centimeters, even if lower than the requirement. Is going to be the best print I can anyway.
dpi print
Most of the photos you will encounter have a value of 72 DPI, this is because that is the value assigned by default by the camera for historical reasons: it was the common resolution for computer displays when digital cameras were introduced to the market, at the time we used 14"-15" CRT monitors. I am not sure I can change it inside my Canon, but it does not matter: most of my photos are to be shown on a computer display and I can change the value (as shown above) when editing for a special print.

Of course, as I told above is specific to photography. For illustration/vector graphics is a different matter, we may talk about at another time if there will be enough interest.

27 January 2014

Nightmare wallpapers

A fan of my photos keep asking for large resolution versions of various pics (I usually put online web-optimized stuff) so I cave in, that's the explanation for me having often posts like this.
The images below are pretty-much made in camera (they were edited with GIMP only for crop, resize, BW conversion and color curves adjustment). The motion blur is created in-camera, with the following recipe (which you can learn looking at EXIF data): put a somewhat long exposure time, 1/10-1/15 of a second, start moving the camera and then press the shutter. The first and the last are made moving the camera on the vertical axis and the second by rotating the camera and zooming at the same time - pretty easy but you will need a bit of exercise to achieve a smooth movement.

nightmare wallpaper
nightmare wallpaper
nightmare wallpaper

22 January 2014

MP4 and Wikipedia

Last week when I read about the request for comments on Wikimedia supporting MP4 video I wasn't at my computer, so I couldn't vote and later forgot about it. My position here is simple: supporting proprietary and encumbered formats can only harm the open web, so Wikipedia and all the other Wikimedia websites endorsing it would be a huge blow for the openness and community, so I strongly oppose. However, I acknowledge a lot of the existing devices produce content in such formats, so I guess a reasonable compromise would be allowing such uploads but then transcode the videos in open formats and serve those for viewing.

Today I was reminded about the issue by a message on some Wikimedia mailing list by a message worded in such way, I can easily label as "pro MP4 spam". It was a good incentive for me to finally answer the RFC/vote.

Still, I am disturbed to see a "product manager" at Wikimedia Foundation campaigning for MP4, especially when the public opinion is strong against it (at this moment it records 251 votes for no support, votes for contributing only, no view and only 132 for full support, so more than 2:1).

08 January 2014

On Red Hat and CentOS joining forces

Red Hat and CentOS joining forces is a really smart move and it potentially is a win-win scenario for all parties involved: Red Hat, CentOS, community, end-users, FOSS. I won't enter into details, since a lot of people are talking about this already.

However, what baffles me is how little people understand the reasoning. So many comments frame it in the old context: a lot of projects need a no-cost Linux solution, and if they can't use a Red Hat derivative, they will use a Debian flavour and later when money will get involved, they will buy support from Canonical instead of Red Hat. Yes, this is a solid reason, but is the old reason. What we see now is something else.

The real reason for Red Hat embracing CentOS now is written plainly everywhere, from the press release "Red Hat is once again extending its leadership in open source innovation by helping to establish a platform well-suited to the needs of open source developers that integrate technologies in and around the operating system" to the FAQ "Red Hat is taking an active role in the CentOS Project to accelerate the development and broaden the reach of projects such as OpenStack by expanding our base."

These days Red Hat is selling not only the established Red Hat Enterprise Linux, but quoting from the same press release, things like "OpenStack, RDO, Gluster, OpenShift Origin, and oVirt" solutions. Compared with Linux, those technologies have so far less traction but comparable potential. Having CentOS as a strong player and supporting all those technologies will make them more popular and increase the chances for a later sale of more than RHEL.

Everything mentioned above is Free Software, so it can be a win-win for all of us. Of course, there are dangers (GNOME 3, I'm looking at you!) but there are safeguars: if the srpms will be easier to get, projects like Scientific Linux will have the life easier too.

02 January 2014

Night wallpapers

For the New Year night I chased some fireworks pictures, but in the wrong place (had to stay near home) so the result was not spectacular (I didn't have high expectations for the fire show in the nearby park). But having the gear with me and noticing the park at night, which do looks different, I turned on them for more pics, which were then posted in the usual places. Receiving a specific request on some social network for high resolution, I follow with a blog post with wallpaper sized versions:

night wallpaper
night wallpaper
night wallpaper
night wallpaper

29 December 2013

Anaglyph

Disclaimer: to properly read the following article, you need to have a pair of red-cyan 3D glasses, otherwise the images won't display as intended. The cheapest one should do it, perhaps even home made. Lacking such glasses, you can generate your own 3D images for different types of glasses.

A few weeks ago I stumbled upon an old article about how James Cameron is/was remaking Titanic in 3D and somewhere in the technical explanation it was mentioned you can do 2D to 3D conversion with Free Software tools, namely the G'MIC plugin for GIMP.

In the late '90ies I received a computer magazine which came with a pair of paper 3D glasses and a CD containing the usual software demos AND a few 3D pictures, it was my first contact with anaglyphs. I still have the glasses, but I use them maybe once every few years, when I remember to search some such images.

Along the time I learned how to properly make such images: two cameras, filters, combine the two images in one. Way too much effort for me. It looks like way too much effort even for some commercial filmmakers...

When I saw the Titanic article, I said to myself 'what the heck, let give G'MIC a try'. AFAIK it is not available in any official Fedora repo, but you can download a binary from the upstream, drop it in the proper folder and go with it. It will survive even a distro update/reinstall. Not a big fan of the 'application inside another application' approach of G'MIC, nor of its duplication of existing GIMP features, that's why didn't have it already installed, but playing is fun.

anaglyph

I don't have the time and patience to learn how to fine-tune the parameters (with hand crafted depth maps you should be able to reach high quality results), nor do I plan to return to anaglyphs any time soon, so I used pretty much the automatic settings. Below are a few pictures I think came decently with no fiddling and automatic settings:
anaglyph

anaglyph

anaglyph

anaglyph

anaglyph

anaglyph

anaglyph

22 November 2013

Mist wallpapers

When I took pictures of the overwhelming mist, it became obvious such minimalist image have a great potential for effective wallpapers: they are simple and won't stand on your way. So here are some for anyone want to use them. Enjoy and don't let their mod depress you :)

mist wallpaper
mist wallpaper
mist wallpaper
mist wallpaper
mist wallpaper
mist wallpaper
mist wallpaper

15 November 2013

Programatica

At first, I received the invitation for the Programatica - Open Source conference for the Fedora community, but since I don't have much to say on this matter, I passed it (I am a Linux desktop user these days and Fedora does not shine there). When the invitation arrived at ProLinux and was difficult to find a speaker, I offered myself for the task.

The conference was mixed, it had 3 tracks: open source projects, open source communities and celebration of 20 years of internet in Romania. The first part, the projects, was about an Arduino clone from Intel (and a Linux distro to match), a Microsoft Azure sales pitch from a local distributor and a F-Droid for FirefoxOS from Ceata. The communities included Ceata (again), "Informatica la castel" summer school, ROSEdu, ProLinux and YATE, which is not quite a community. The last track brought panels and talks with some of the internet pioneers in the country.

My talk was a short one and quite generic (slides here): in an age when so much money are invested in Linux and Open Source and so much money are drawn from them, there is still a need for the community? We believe so.
programatica