M.Chapman

Verified
  • Content count

    1,125
  • Joined

  • Last visited

Everything posted by M.Chapman

  1. If you're happy with their T&Cs you could upload a full size image for free here, then copy and paste the link the site provides to the uploaded image into your Alamy forum posting. You should then get more reliable comments on your image. Also suggest making a note of the delete image link for future reference, or set an expiration period. Mark
  2. I couldn't help chuckling when I read this sentence in the job advert... We are faced with a fascinating, intellectually demanding challenge of understanding the behaviour of search engines, crawling and indexing and understanding customer behaviour on our site. We are looking for a bright person to help us meet this challenge. Will they be starting with gaining an understanding of how their own search/ranking engine works?? Hopefully the search engine is now finally working the way Alamy intended when they introduced the algorithm changes over 6 months ago, but it's taken a heck of a long time to get there. Still, looking on the bright side, it sounds like they are still investing in making improvements. Nevertheless I hope any future search engine revisions are more thoroughly tested before going live. Note to new recruit - Action 1 - Set up a parallel test system so that changes can be fully tested offline. (Sorry, am I being a bit sceptical?) Mark (anxiously awaiting impact of latest changes as highly rated captions were very good for me)
  3. Happy fun images...and a few random ones

    Seems to be a theme to the "random" images hereabouts? Mark (just going with the flow)
  4. The maximum size I can access on Flickr is 2048 x 1361. It's not easy to make a judgement without being able to see the image at 100% (4912 x 3264) that was uploaded to Alamy. Looking at the ~40% view I agree the boat in the foreground looks softer than the buildings in the background. I think at 100% the boat would probably look decidedly soft. Mark
  5. AIM issues - An update please

    I often find when trying to add a tag to multiple images it fails because at least one of the images has reached 50 tags. But there's no easy way to find out which one. I 100% support Stokie's idea "Some sort of feature whereby when the 'exceeded 50 tags' message is received the new tags are only added to those images that haven't reached 50 tags." The other thing I find confusing when working with multiple images is the way that a tag only turns black if it's a tag in all images, or if it's a supertag in all images. If it's a tag in some images and a supertag in others it's grey. I then click on a tag that's grey thinking I need to add it to the other images, only to be given a message about it already being a supertag and the tag isn't needed, or something like that. Not sure what the solution is here (maybe I just need a new brain!). I also find the jumbled list of tags very hard to read. Tag deletion is also difficult. Firstly you have to find the tag, then click very precisely. I suppose, in conclusion, my feeling is that the additional features these clever tag "buttons" provide are just too fiddly/complicated unless you're only working on very few images at a time. On balance I think I'd prefer a simple text box list of tags and supertags too, just like the old AIM. But, if the cut and paste was to work properly and not be saved straightaway when pasting, and Stokie's request was implemented I could work around it.
  6. Slow month?

    Hi Marb, Just my two cents worth.... If someone wanted an image for the following (your suggestions) Vintage steam engine articles Articles on restoration Articles on rust Articles on abandoned industrial buildings etc..etc..etc.. Why would they want a black and white image "HDR styled" image? Wouldn't they be looking for a more "factual" image that illustrated any of those more effectively? It could however perhaps sell for an article in a photo magazine about black and white or HDR processing? Suggest deleting keywords like rivet, rivets, wheel, dial, dials, brick, pump from this image to reduce false hits which will damage your ranking. I like the image, but it's more "arty" than editorial. That's not to say it won't sell on Alamy, it's just that the chances are much, much lower. I've got quite a few "arty" images on Alamy myself, but in general they are a waste of time (I'm still learning). So I'm now trying to sell those elsewhere. This on the other hand has recently sold on Alamy, to illustrate rust and corrosion in an educational book. Mark
  7. English Term please

    I would also add circular, round, conical, thatched roof, woodland shelter, woodland refuge, forest shelter, forest refuge Mark
  8. Alamy has a "virtuous circle". If you put your best images here, you're more likely to get sales. If you get more sales your Alamy Rank will improve, this will lead to your images coming higher up in searches. This will lead to more views and, if the images are good enough, more sales. Success can grow on Alamy. Similarly, failure to put your best stuff here can lead to a "vicious circle" of decline. No sales, declining ranking, fewer views, disappointing sales over 2 years. Give Alamy your best shots, keyword them well, and you may be less disappointed. Mark
  9. IM...

    If both images are on sale and have been for 24 hours, you can use OR in the search box with the Alamy refs. For example seach for AB1234 OR FG1234. It's a real shame this doesn't work until both images are already on sale. Mark
  10. Hi Jeff No need to feel guilty. I took it as a challenge to see if those bars could be removed, and learnt quite a bit on the way. The app I found that removes lens correction data will also be quite useful to me as it recovers so many pixels which can often be used in shots that don't contain architectural details. I'll look forward to the scones, jam and tea. I might even bring some cream Mark
  11. Interesting. I note that there's also quite a lot of discussion about over-exposure warning stripes too. Looking at the images above I note that the stripes are misplaced to the right, relative to the highly exposed areas. That's weird... If the problem was in the sensor (e.g. overflow) shouldn't it be more perfectly synchronised with the light areas? Is it possible it's a rogue bit of "over-exposure warning" code in the firmware that's leaving its footprint? Mark
  12. Glencoe Trademarked!

    I imagine NTEW and NTS may have noticed what happens when such letters get posted on social media... The power of social media and the damage it can do to your brand and reputation is becoming quite significant. Is there a little ® on the the road signs when you enter Glencoe? I think not. Mark
  13. Hi Jeff, A minor refinement that allows downsizing by just 10% (instead of 50%). After opening the PSD* in Photoshop, do the following; 1) Use Image>Resize>Canvas Size... to chop 4 pixels off the right-hand edge of a horizontal format image so the image becomes 5,460 pixels wide. i.e. divides exactly by 10 2) Use Image>Resize>Image Size... to reduce pixel size to 90% using nearest neighbour algorithm (this will throw away 10% of the pixels (1 in 10 columns and rows are discarded. The PSE 90% resize removes pixels in the 5th, 15th, 25th etc... columns, and luckily these are exactly the columns which contain the dark band defects. 3) Note that some white defects remain - look at the distant building skyline. So some cloning may still be needed. The above works in PSE, I hope it's the same in CS6. Here's before (right) and after (left) 100% crops from the latest DNG you posted showing almost perfect removal with just a 10% downsize. *I used On 1 to create the PSD file without any lens distortion correction from your DNG. It's possible other RAW convertors may work, but I got the best result with On 1. If you do use a different convertor the columns that the defect bars appear in may shift by +/-1 pixel, so the resize canvas process would need altering. I did some checking to see where the lens distortion correction data used by ACR/LR is stored. It's not an Adobe profile, it's supplied by the camera manufacturer and is stored inside the DNG file, hidden in the OpCodeList parameters. I don't have any tools to edit these, but did find an app (for Mac) which can remove them, see https://itunes.apple.com/us/app/dng-cleaner/id1205138449?mt=12 . When I tried it, it worked fine and when I opened the file in LR the lens correction was removed, but defective pixel bands weren't pure 1 pixel wide. There was some weird "bleeding" into adjacent pixels (see below) probably due to a different demosaic algorithm in LR. So that's no good, which is a shame, as it would have provided a fix that didn't need another RAW convertor. PS. Interestingly it looks like On 1's RAW convertor is quite good. A good balance between detail and noise which I struggled to beat in LR. Mark
  14. Hi Jeff, I found an easier fix for the defect bands. I looked at your latest (RAW) DNG and it's now clear that lens distortion correction is being applied by ACR or LR when I open the file. LR says This raw file contains a built-in lens profile for correcting distortion, chromatic aberration and vignetting. The profile has already been applied automatically to this image. This cannot easily be turned OFF in LR or ACR. So, I downloaded On1 RAW (a dumber image editor) which doesn't carry out automatic lens distortion and opened your DNG in that. Bingo. Image distortion correction is no longer applied. The image shows loads of vignetting and barrel distortion (it's a pretty poor lens/camera), but now the defect bands are perfectly vertical and only 1 pixel wide (if on white background) and (where they appear) they only occur on every 10th column of pixels. So now all we need to do is throw away those columns of pixels. I used On 1 to export as a 16 bit PSD and then opened in PS. I then used Image>Resize>Pixels 50%* using nearest neighbour and the defect bars are completely (and perfectly) eliminated! The image now needs fixing for vignetting and distortion, but that's not too tricky. It will also need slight upsizing to meet minimum Alamy size (17MB) *In theory it should be possible to downsize by just 10%, but I had trouble synchronising the columns of pixels that are discarded with the defect bars. I'll investigate this further. Mark
  15. Hi Jeff, I did some more playing with the DNG and got some encouraging results. 1) Remove lens distortion correction in LR so that the bars all appear vertical (and max 2 pixels wide). Picture looks a bit "fish eye", drone camera lens obviously suffers from massive barrel distortion. :-( 2) Export to Photoshop as PSD or TIFF 3) In PS. Image resize to exactly 50% of width and height in pixels using nearest neighbour algorithm (makes bars only 1 pixel wide) 4) Duplicate background layer (becomes top layer) 5) Apply Custom filter 1 -1 1 with scale 1, offset -10 to top layer. This will turn the high contrast vertical black bars white. (It's basically an invert and sharpen filter for vertical edges). 6) Apply threshold of 249 to new layer (so <249 = black and >249 = white) 7) Select all white areas on new layer (magic wand selection tool, tolerance 1, non-contiguous). This effectively makes a mask. 8) Hide top layer and select background layer to work on 9) Hide selection mask outline (cntrl H) so you can see what's going on 10) Apply Filter>Noise>Dust and scratches radius 2 threshold 1. Boom! the bars pretty much disappear 11) Clear selection (deselect) 12) Delete top layer 13) Reapply distortion correction 14) Upsize image to meet minimum Alamy size requirement (17MB). Maybe the above can be stored as a PS action? NB. Although it's a partially automatic process it's still far from perfect and would still fail QC without some manual cloning. It might work better if starting from a true RAW image. Here's a screenshot before (right) and after (left) step 10. Mark
  16. I'm certainly not giving mine away! It was expensive and I like it too much.
  17. Hi Jeff, If you open the unprocessed RAW image in CS6 then yes, you might* be able to reverse the automatic distortion correction by adjusting using the lens distortion correction tab and dragging the slider back towards barrel distortion (I found -37 was about right). Try to get the black bars near the corner of the frame to turn back into very narrow (ideally single pixel) perfectly vertical stripes. (It may not be possible to get single pixel width, depending on where the fault is occurring and the demosaic algorithm) *There's no guarantee you'll be able to get a perfect cancellation of the automatic distortion correction since the automatic correction may use a more complex distortion adjustment (based on parameters in the raw file, or in Adobe's lens profile model). You could try turning the any automatic lens distortion correction off, if that option is available. But with my camera that requires editing the EXIF as that's were the parameters are stored. If you can get a good cancellation so the bars are only a pixel wide and not smeared, then try the noise and dust removal filter in CS6 (I think it's got the same filter PSE has). Try Filter > Noise > Dust &Scratches. Suggest setting a high threshold and the radius as small as you can get away with. On PSE it removes dust and scratches and does a content aware fill, and yes, it works on the entire image. But.... be careful to check that other narrow dark objects (telegraph poles and wires etc.) haven't also been removed! You may want to use in combination with a selection brush to avoid collateral damage. Again, if you can post an unprocessed RAW image, I'm happy to experiment. Mark
  18. Supertags?

    That sounds awfully familiar. There was a long thread on supertags demoting images in February. See here http://discussion.alamy.com/topic/7118-proof-“supertags”-are-not-working-properly-in-fact-they-make-image-position-worse/. I thought it was fixed though?? Maybe it's come back? Or... it never really went away? Mark
  19. I finally got fed up with Microsoft forcing updates on me that often seemed to cause problems. So I bit the bullet and bought a secondhand MacBook Pro. After feeling like a "fish out of water" for a while, I've now grown to love it. It just works. The user interface is, more consistent than Windows 10, although I don't find Finder as intuitive as Windows File Explorer. But, in almost 2 years I have had NO problems whatsoever, even with major updates from Yosemite to El Capitan to Sierra. My productivity has shot up. I still run Windows, but only when I have to (using Parallels to run a copy of Windows 7 as a Virtual Machine). Mac OS seems to manage memory and tasks way more efficiently than Windows. It's got to the point I rarely bother to close anything. Battery life is superb, performance is just as quick or quicker. Windows seemed to need constant "maintenance" to keep it working properly, and periodic reinstalls. Mark
  20. No more PMs

    +1 I recently used it to help a new contributor understand how to pass Alamy QC. We exchanged real email addresses and then RAW files to illustrate the problem. I didn't use the feature very often, but it will be sadly missed, especially as I'm reluctant to post my real email address anywhere in public view (to avoid spam). I hope Alamy will reconsider. Mark
  21. No more PMs

    IOC. That's a REAL shame. Will the Forum be next? Too much negativity around here methinks... and this won't help
  22. No more PMs

    Was there an announcement?
  23. Wim, I've downloaded the DNG. I believe the stripes would be parallel if we could see the raw sensor data before automatic lens distortion and CA correction that has been applied. If I apply -37 manual lens distortion correction in LR (in an attempt to cancel the correction already applied) the bars become vertical and parallel. They are still slightly "smeared" due to the double distortion correction. I think, if we could see the original data they would be perfect 1 pixel wide dark vertical lines that occur only in clipped highlight areas. Looks like a sensor or data processing problem in the camera to me, and nothing to do with the SD card. Jeff, The DNG file you posted looks like it was produced in LR or with ACR (it has all sorts of corrections applied highlights, vibrance, etc.) do you have a *real* RAW file from the camera that hasn't been processed? If you have it's sometimes possible to turn off distortion correction (by editing the EXIF settings) or by using DXO optics to convert the RAW. This would then leave the dark bard in their "purist" form, potentially making them much easier to remove, for example using the PS automatic dust and scratch filter with high threshold. Mark
  24. Wim, It's not moire. Look at the sides of the white vehicles in the picture in the original posting. It looks more like some weird highlight overflow wraparound binning problem in the sensor. The slight distortion of the bands (they don't appear to line up perfectly with the sensor pixel grid) maybe caused by some in-camera lens distortion correction? Jeff, If you can post a RAW version of that image somewhere, then we can take a closer look. If all the bars are a single and identical colour and ONLY occur where the image should be white, then some of the global selection tools might work. Mark
  25. Certainly can't think of an easy way, and with $/image so low, it's not worth the time - unless some of the images are truly unique. Might be easier if the bars are in the sky, but still overly time consuming. IMHO. Why not submit one captioned and keyworded as "Image showing digital camera sensor defect"? Although it might get a QC fail on the way! Mark