Tips on improving these photos

Hi,

I have several photos that were taken by a point and shoot and I am looking for some tips in order to improve the quality. I just want to be sure that I'm on the right track. I only have these photos and not the raw files so I realize that a lot of data is missing and I have to make the best of it. I use photoshop CS4.

Could the experts on here have a look at them and try to improve them? I would love to know how it's done. I have completed one of them and would like to see what you guys think of the job.

http://backup.cambridgeincolour.com/...3.original.jpg

My Revised copy of the above: What do you think?

http://backup.cambridgeincolour.com/...7/original.jpg

Also help needed on these two:

http://backup.cambridgeincolour.com/...5/original.jpg

http://backup.cambridgeincolour.com/...4/original.jpg

I would really appreciate any suggestions. Am on the right track here?

Hi James,

I think that the fundamental issue is that your in a situation that requires a lot of dynamic range (there are very bright areas and relatively dark areas) (ie "a very contrasty scene") - and yet in using a point & shoot camera (with a very limited dynamic range) you quickly find yourself in a situation where the camera uses most of it's range exposing the highlights correctly ...

... and then you find yourself having to dig out shadow detail, only to find a lot of noise there as well.

In an "ideal world" I would do as many of the following as possible ...

- Shoot with an SLR camera

- Shoot at a time where the light outside isn't a lot brighter than the light inside

- Use a flash and diffuser inside

- Consider mounting the camera on a tripod and shooting a series of shots in 2 EV steps - and combine these into an HDR photo (High Dynamic Range).

Does this help?

I'm far from expert but as Colin has already mentioned the trouble is too wide a range to capture in 1. Even a raw file wouldn't be able to encompass the bright window whilst capturing the room at optimum brightness since the problem is the actual range of the camera. SLRs are better as Colin said but even they fall short in dim interior/strong sunlight outside scenes.

Perhaps another suggestion which is similar to HDR is manually blending images, this way you may or may not require a tripod depending on situation, workflow etc. Failing recapturing if you ever get stuck and only have the imperfect image to work with and cannot recapture again for whatever reason then it is possible to save it, but it's never as good as getting it right to start with. After all the end product depends on the raw materials, put sub standard ingredients in and it will affect the outcome no matter how good a chef you are it's just not the same as starting with better quality input.

Oh obviously out of the 2 types of imperfect exposures only 1 can be saved. If it's exposed for shadows then it might be a lost cause because there is no way of recovering highlights. Obviously in this scene you could just fake a window if you really had to but in most situations the pic is pretty much unsalvagable if it's blown. If the hightlights are not blown (or relatively small proportion of the highlights are) then you can sort of save it but it will drag the image noise up as you can see in your own first example. With a couple of passes and tweaking you can brighten a fair bit, eg.

http://backup.cambridgeincolour.com/...43original.jpg

you can brighten a fair bit bt as you can see there is detail loss, artifacts (like halos from trying to recover detail after blitzing noise) and saturation washed out due to aggressive curves. Duplicated layer. Denoised and softened and curves adjust, then sharpened to recover over soft effect, did these in small steps reopening in acr to repeat at different stages by different amounts etc. Finally when noise under control and brightness ok vibrancy up and then sponge tool (saturate) on selected bits to get rid of washed out look. Finally added layer mask to original untweaked image with the window masked off so the window is identical to original (otherwise would be blown out due to aggressive brightening).

James,

For what it's worth, I am not sure that your cloning out of the plant in the window far left of the first photo has been that successful. This area jumped out at me as having been "interfered with" because the blinds are "broken". I think you need to do the cloning in much smaller steps, ie using a very fine brush, perhaps concentrating on the thick horizontal bar first. Well, or don't clone the plant out at all. I think you will find that the plant helps stop the eye wandering out of the photo at that point. But perhaps the client wanted it removed?

I did have a go at your last photo - the dark one. Masking out the window as Davey suggested earlier. I used curves to brighten the darkest areas, but ended up with a lot of shadow noise. I'm no expert at dealing with that, so have given up, sorry.

Graham

A few things,

I agree, that the plant should stay intact. I will keep it - it was a rush job and i noticed the same things you did with regards to the broken blinds.

I am also trying the deal with the darker photo. It's a tricky one because of the really dark areas. What I'm basically looking for is the technique in dealing with them becauase then I can do it myself.

Anyone else want to have a go at them?

Would the experts suggest cropping the photos into a 2:3 ratio ie 4 x 6? Is that more pleasing to the eye and on the www? What is the standard?

Quote:

---

Originally Posted by james

Would the experts suggest cropping the photos into a 2:3 ratio ie 4 x 6? Is that more pleasing to the eye and on the www? What is the standard?

---

Sorry, we're fresh out of experts, just little old me, and I'm more old than little :o

I wouldn't say there's a standard, I feel each pic should be cropped to a ratio that suits it compositionally, for example, I very rarely 'do' square pictures, but for my recent moon pics (in another thread) it was the obvious choice.

In this particular instance, I'd say the subject would not suit 3:2 as most rooms aren't that shape (wall-wise), so it could look claustrophobic.

Cheers,

I really like to result of Davey's edit to the picture and would love to know in detail how this was done. I have tried following his instructions but without more precise details, I don't think I'll be able to get a similar result.

Out of curiosity James, why are you shooting these with a P&S? What happened to your Nikon SLR?

Hi Colin,

I didn't actually shoot these photos, I'm just trying to improve them for someone and learn while in the process.

Quote:

---

Originally Posted by dendrophile

but ended up with a lot of shadow noise. I'm no expert at dealing with that, so have given up, sorry.

---

But I have just found Neat Image... which does a great job at removing noise. Like it! With the demo version I have been able to control the noise in the shadow areas somewhat. This is what I got:

http://backup.cambridgeincolour.com/...c/original.jpg

What I have done here is (in Photoshop):

1. Use the neat image plugin to remove noise. It wanted to use 70% in Y, so I let it!

2. Used the polygon lasso tool to select the window area. I then feathered the selection by 2 pixels and inverted the selection.

3. Used curves to brighten the photo.

4. Because the photo is basically monochrome, I used vibrance at +10 and saturation at +10 to put a little colour back.

5. Job done.

If you crank the brightness up to stupid levels (say, setting the middle point to 2.5 using Levels) you get enough of the bouquet of flowers visible to allow you to at least make out the leaves. Possibly you could this as a template and paint on some colour to this area. I've done something simple, copying this area to a new layer, giving it an opacity of 44%, and then used a soft brush to paint onto a layer mask to blend the edges a little better. You can see the extra noise in the flowers on my version. Not totally happy with it, but getting a little foreground interest into the photo seemed to improve it.

HTH,
Graham

Noise ninja is also worth looking at. My brother uses neat image (as does my dad I think) and I like it although it destroys a lot of detail at times. Also pshop alone can give same results in many situations but it's fiddling and for batch processing it's not ideal (unless it's unifom then actions are helpful). As for noise ninja I've never used it but seen it as a few friends use it and say it's pretty much the same as neat image (sometimes better sometimes not quite as nice).

Detailed workflow is as follows with a few extra suggestions I didn't do, if you still struggle understanding anything then give me a shout and can go into more detail for whatever step. Probably not the best workflow and I'm sure many others have suggestions that may be more suited. Whatever fits you best.

1. Opened, converted to 16bit and saved as tif. 16bit to be on safe side with regards to aggressive colour etc edits but not essential. Any lossless format would do but I intend to open in ACR (adobe camera raw) so that narrows it down to tif, I usually work in psd (or png or targa depending what I'm doing and sometimes tif).

2. Save and reopen, now it's opened by acr. Note: you need to have open all tif files selected in acr prefs.

3. Turn luma and chroma noise reduction to 100. No sharpening keep it soft. Open in pshop proper.

4. Filter>noise>despeckle and then filter>noise>reduce noise (settings strength 10, no sharp, preserve detail 10%, chroma 100 (even though acr took care of it mostly).

5. Save now as will be using this one for the window.

6. Now for brightness: curves adjust, pull up midtones. Preset closest to what you want is "lighten" but might have to drag around a little bit. Tweak curves some more to get brightenss right. Can save (but not over window one) and reopen with acr if that is quicker/easier for you.

Not usual workflow for most going back to acr including for myself as only use it for initial raw developing mainly. But sometimes I go back to it for certain things pshop doesn't do the same or it's quicker in acr. For instance in this case I prefer exposure tools along with clarity slider as set out in acr even though can get access in pshop to same features just for speed of a quick dirty edit. Get brightness about right then adjust contrast (s shaped curve), clarity, exposure, black clip point etc until looks "right" again. Also dodge and burn tools might be good but it's longer task. Ignore how the window looks here.

7. Open window file seperately and turn recovery up to max. Copy to new layer main image you are working on. Mask off window by whatever method you find easiest (box, lasso, pen, wand etc)and feather a little too maybe (I didn't but was quick dirty job). Apply new layer mask and it should auto paint all but the window out. I didn't but guess you could make selection a path and "stroke" it with a 50% black brush to blend the window edges even more, depends how far you wanna go really. When it looks natural flatten image.

8. Now all that should be a main problem is colour is washed out so increase vibrancy. Also sponge tool set to saturate helps, and vibrancy/sat layer adjustment with mask applied might be worth it depending what you want (I didn't do this but would be useful since I noticed the main thing you don't want to re-colour too much is the walls as go yellow so make the global adjustment then paint out on the layer mask). Whatever is quickest for you that works.

9. Now it all looks good apart from it's way soft so wack on whatever sharp you want. Again uplicate layer and sharpen then layer mask off areas you don't want any sharp on like the walls. Should be looking better now.

That's a little more than what I did, and you could do more but it depends on what it's for. I obviously only did rushed job but you will be wanting to do a better job I'm presuming. If it was for a worthy cause then really going to town on tweaking and spending an hour obviously isn't a bad thing but it's not always worth it or necessary. Spending 2 days on tweaking a passport photo is probably overkill (hmm not if you'd seen my face hehehe) but a 3 min rush job on a grainy but only surviving pic of your friends deceased mother isn't really ideal if you know what I'm saying.

Hopefully my method might help, but it might also get you into bad habits workflow wise, sorry! Obviously others might be able to suggest refinements or easier ways. As the saying goes there is a difference between a short cut and cutting corners. Shortcuts are always good way of achieving same thing quickly, cutting corners leads to lower quality results but it's a necessary evil at times. Obviously if you have 300 pictures to process in 2 days then my method isn't ideal (unless you like RSI and sleep deprivation).