OPINION OF THE COURT
Andrias, J.The verdict convicting defendant of second-degree murder at his third trial was not against the weight of the evidence (see People v Danielson, 9 NY3d 342, 348-349 [2007]). The issue to be determined is whether the trial court committed reversible error when it precluded a defense expert from testifying about the effect of “wéapon focus” on eyewitness identifications without conducting a Frye hearing (Frye v United States, 293 F 1013 [DC Cir 1923]), or when it ruled that if the expert testified about “the effect of postevent information on accuracy of identification,” the People could elicit evidence that the identifying witnesses cooperated with police to produce a composite sketch, which resembled defendant.
*101 We hold that under the particular circumstances of this case, the proposed expert testimony as to weapon focus would have had little relevance and its exclusion did not prejudicially deprive defendant of a fair trial. Moreover, the testimony regarding the preparation of the composite sketch would have been admitted for the limited purpose of allowing the People to address a potentially misleading impression that would have been created by the defense identification expert’s proposed postevent information testimony, and the conditional ruling was not unduly prejudicial to defendant, given the limiting instruction proposed by the court. Therefore, we affirm the judgment.
On June 15, 1991, livery cab driver Joaquin Liriano was stabbed to death. A number of people witnessed the attack, and within days four of them collaborated on a composite sketch of the perpetrator. In 1993, defendant was deemed a suspect in the homicide after he was arrested for an unrelated burglary and an officer thought that he resembled the composite sketch. The police were unable to locate any of the witnesses and the homicide case remained dormant until April 1998 when defendant was arrested for another burglary and the police again concluded that he resembled the composite sketch.
The authorities then located the witnesses who contributed to the composite sketch and another who did not come forward until 1998. R.R, who saw the perpetrator from the street, positively identified defendant in a photo array and lineup as the perpetrator. T.F., who saw the perpetrator at a distance of 20 to 25 feet through a window in his apartment, viewed the photo array and noted that defendant’s photograph was a “very close, if not exact match.” J.G., who was with T.F. at the time of the attack, noted that defendant’s photograph looked “similar” to the perpetrator. The fourth and fifth witnesses (L.G. and S.G.) were unable to identify defendant. No forensic or physical evidence tied defendant to the stabbing.
Defendant’s first trial ended in a hung jury. He was found guilty at his second trial. This Court affirmed (28 AD3d 318 [2006]), but the conviction was reversed by the Court of Appeals (8 NY3d 449 [2007]) (LeGrand I) and a new trial ordered on the ground that the trial court erred when it precluded the defense expert’s proposed testimony regarding the lack of correlation between confidence and accuracy of identification, confidence malleability, and the effect of postevent information on identification accuracy, the underlying principles of which were *102generally accepted by the relevant scientific community* The Court of Appeals found that testimony as to the psychological phenomenon of weapon focus was properly excluded because there was insufficient evidence to confirm that the principles underlying the proposed testimony were generally accepted in the relevant scientific community at that time.
In so ruling, the Court of Appeals established a two-stage inquiry for considering a motion to admit expert testimony. In the first stage, the court decides whether the case “turns on the accuracy of eyewitness identifications and there is little or no corroborating evidence connecting the defendant to the crime” (8 NY3d at 452). In the second stage, a court must consider whether the proposed “testimony is (1) relevant to the witness’s identification of defendant, (2) based on principles that are generally accepted within the relevant scientific community [which may require a Frye hearing], (3) proffered by a qualified expert and (4) on a topic beyond the ken of the average juror” (id.).
At the third trial, defense counsel informed the court that she would not be calling her identification expert to testify in light of the court’s rulings that expert testimony as to the effect of weapon focus would not be allowed and that testimony on postevent information would open the door to testimony about the composite sketch. RE and J.G. identified defendant as the assailant; T.F. was too ill to attend and his testimony from the prior trials in which he identified defendant was read into the record.
Defendant argues that he established a sufficient basis for at least conducting a Frye hearing on the issue of the effect of weapon focus and that the court’s ruling prejudicially deprived him of the opportunity to elicit a form of valuable evidence. This conclusion is not supported by the record.
“As a general rule, it is an abuse of discretion to deny a motion for expert testimony on eyewitness identifications in a case that depends solely on the *103accuracy of eyewitness testimony if there is no corroborating evidence connecting the defendant to the commission of the charged crime and the proposed testimony satisfies the general criteria for the admissibility of expert proof. [The Court of Appeals’] concern in recent years has arisen from psychological studies that have addressed the potential for misidentification when a person observes an assailant — usually a stranger — for the first time in a highly stressful environment” (.People v Muhammad, 17 NY3d 532, 545-546 [2011] [citations omitted]).
However, even where expert testimony should be admitted, the court is still obliged to exercise its discretion with regard to the relevance and scope of the testimony and not all categories of such testimony are applicable or relevant in every case (see People v Santiago, 17 NY3d 661 [2011]).
“ ‘Weapon focus’ is the phenomenon which occurs when, during the course of a crime, a witness is exposed to a weapon, and the witness focuses his or her attention on the weapon and not on the perpetrator’s face, which [allegedly] impairs the ability of the witness to make a subsequent identification of the perpetrator” (People v Banks, 16 Misc 3d 929, 931 n 4 [Westchester County Ct 2007]).
While the testimony of RE, T.F. and J.G. establishes that they did take notice of the knife when they saw defendant stab the cab driver, none of them were close enough to defendant to be threatened by it, and thus were not within the zone of danger. T.F. and J.G. were inside their apartment and RE moved to a position of safety behind a vehicle, where he continued to watch the attack on the cab driver unfold.
Significantly, the identifying witnesses’ observations of defendant were not limited to the point in time when the cab driver was stabbed. RE testified that after the stabbing, defendant walked away and then returned to retrieve a “multicolor African” type bag from inside the cab. At that time, RE had an unobstructed view of defendant and got a good look at him. RE saw defendant’s face and watched him calmly walk towards Morningside Fark. When the cab driver was on the ground, J.G. saw defendant walk around to the other side of the cab, pick up the multicolored bag and walk away “like nothing happened.” As she watched this going on, she was able to see *104defendant’s face. T.F. also observed defendant grab a multicolored bag from the cab and casually walk away. S.G. did not identify defendant, but her 911 call was offered to show that the incident lasted about two minutes.
Further, unlike People v Abney (13 NY3d 251 [2009]), where the victim did not describe the assailant as possessing any unusual or distinctive features or physical characteristics, R.E testified that defendant’s face had cheekbones like Mike Tyson, and a moustache and sideburns “like Elvis.” T.F. and J.G. were struck by the fact that defendant did not appear angry and had a blank expression on his face. J.G. also explained at trial that she wrote “similar” on the photo array because “it looks exactly the same as what I saw outside the window.” T.F. felt that the photo in the array was a “strong match,” but it was a “black and white photo, photocop[ied] with wavy lines in it so it’s not a crystal clear sharp black and white photograph.” T.F. explained that there was a “profound difference” between seeing a “degraded photocopy of a person” and seeing that person at the time of the incident and in the courtroom.
In light of this testimony, which establishes that the identifying witnesses got a clear look at defendant as he retrieved the bag and casually left the scene, at which time the knife was no longer a point of focus, the proposed expert testimony as to the weapon focus effect had little relevance and its exclusion was at most harmless error which did not prejudicially deprive defendant of the opportunity to elicit a form of valuable evidence. In this regard, we note that defendant was still free to produce expert testimony regarding the lack of correlation between confidence and accuracy of identification, confidence malleability, and the effect of postevent information on identification accuracy.
Defendant also argues that the trial court effectively precluded his identification expert from testifying when it ruled that his proposed testimony on postevent information, authorized by the Court of Appeals in LeGrand I, would open the door to testimony about the composite sketch. Defendant maintains that the trial court’s holding that evidence relating to the preparation of the sketch was admissible to show “unintentional falseness” is contrary to the holding in People v Maldonado (97 NY2d 522 [2002]) that such evidence may not be admitted to counteract evidence that a witness’s identification was the result of mistake or confusion. While the trial court would not permit the actual sketch to come into evidence and proposed a limiting *105instruction, defendant argues that under Maldonado, the mere mention of the sketch would have been impermissibly prejudicial.
In Maldonado, the victim identified the defendant as one of the assailants in an attempted murder/robbery. However, on cross-examination, the victim was shown a photograph of the defendant’s brother and identified him as that assailant. The defense also elicited testimony from a detective that fingerprints at the scene did not match the defendant’s and that the detective never showed the victim a photograph of a third assailant. The prosecution then sought to introduce the composite sketch prepared by the victim and a police artist, on the ground that the defense had opened the door by creating the impression that the victim’s identification was incorrect. The trial court, over defense objection, admitted the sketch and gave no limiting instruction as to its use or purpose. In summation, the prosecutor emphasized the importance of the sketch.
The Court of Appeals reversed, holding that a composite sketch may not be admitted to counteract evidence which casts doubt on the reliability of a witness’s identification, because it impermissibly bolsters the witness’s testimony (97 NY2d at 529). The Court explained that, especially where the identity of an assailant is in dispute and the proof rests entirely on identification, “[m]ere mention at trial that an identifying witness cooperated with a police artist to produce a composite sketch is, under most circumstances, impermissibly prejudicial” (id. at 528 [emphasis added]). The Court did carve out a limited exception to the hearsay rule barring the admission of a sketch, holding that it “may be admissible as a prior consistent statement where the testimony of an identifying witness is assailed as a recent fabrication” (id. at 529) to confirm the identification with “proof of declarations of the same tenor before the motive to falsify existed” (id.). The Court found that the exception did not apply in Maldonado because the defense suggested that the identification was a mistake, rather than a recent fabrication. Nor did the cross-examination of the detective open the door since the inquiry “merely completed the account brought out by the prosecutor, and did not carry an implication of fraud or bad faith” (id. at 530).
As the trial court noted, the situation here differs from Maldonado. The court did not rule that the sketch itself would be admitted. Nor did the court in any way preclude the proposed expert testimony as to lack of correlation between confidence *106and accuracy of identification or confidence malleability. Rather, the court ruled that if the defense intended to have its expert testify that the witnesses’ identification was tainted by postevent information (in the form of the photo array), evidence as to the preparation of the sketch would be allowed so the jury would have a true picture of what transpired, i.e. that shortly after the event the witnesses could recall and relate their observations of defendant. Unlike Maldonado, there were no conflicting identifications that had to be explained away and the court made clear that it would give a limiting instruction as to the use of the testimony, which was not being introduced to prove defendant’s guilt.
A party may open the door to otherwise inadmissible evidence by creating a misleading impression before the jury (see People v Massie, 2 NY3d 179, 180 [2004]; People v Lindsay, 42 NY2d 9, 12 [1977]; People v Heckstall, 45 AD3d 907, 909 [2007], lv denied 10 NY3d 766 [2008]). In these and similar cases, the witness had multiple opportunities to identify the accused, and the defense was unfairly selective in an effort to suggest that the witness was unable to make an identification. Here too, had the defense expert been permitted to testify about postevent information in order to discredit the eyewitness’s in-court identification of defendant, without the People being able to introduce evidence relating to the preparation of the sketch, it would have created the misleading impression that the eyewitnesses were unable to identify the assailant without seeing the photos (see People v Lopez, 9 AD3d 692, 694 [2004]).
Accordingly, the trial court did not err when it ruled that in order to avoid giving the jury an incomplete recitation of the events leading to defendant’s identification at the photo array, and to the extent that the defense intended to argue that the array caused the witnesses to identify defendant, testimony could be elicited that the witnesses met with an artist shortly after the crime, gave a description of the perpetrator, from which a composite sketch was created, and then years later the witnesses viewed the photo array. As the trial court stated, this would balance things “by giving the jury an understanding of what each of the witnesses’ bases for the identification was as to whether they fabricated, either subliminally or induced to do it by the picture of him, or whether this goes back.” The sketch itself would not have been admitted into evidence and a limiting instruction would have been given that the testimony as to the sketch could be used only to evaluate the expert’s opinion that *107the eyewitnesses’ in-court identifications were caused by having viewed defendant in the photo array years after the crime occurred.
We note further that in Santiago, when the Court of Appeals ruled that “Supreme Court should also have given more adequate consideration to whether the proposed testimony concerning exposure time, lineup fairness, the forgetting curve, and simultaneous versus sequential lineups was relevant . . . and beyond the ken of the average juror, and if necessary held a Frye hearing,” it stated in footnote 4: “We note that the fact that the victim assisted in the creation of an artist’s sketch of her attacker does not render the expert testimony irrelevant” (Santiago, 17 NY3d at 672). While the eyewitnesses’ assistance in the creation of a sketch does not render the expert testimony irrelevant, we believe that its existence may be disclosed where it has a direct impact on the value of the expert’s testimony on a particular point. Here, with the appropriate limiting instruction, it would not be unduly prejudicial to question the defense expert as to whether his opinion on the effect of postevent information would change based on evidence that the eyewitnesses had contributed to the preparation of a composite sketch of the perpetrator within days of the crime and before they saw the photo array.
Accordingly, the judgment of the Supreme Court, New York County (Lewis Bart Stone, J.), rendered March 3, 2009, convicting defendant, after a jury trial, of murder in the second degree, and sentencing him to a term of 25 years to life, should be affirmed.
Andelas, J.E, Sweeny, Renwick, Freedman and ManzanetDaniels, JJ., concur.
Judgment, Supreme Court, New York County (Lewis Bart Stone, J.), rendered March 3, 2009, affirmed.
“Confidence-accuracy correlation refers to the relation between the accuracy of an eyewitness’s identification and the confidence that [the] eyewitness expresses in the identification.” “Postevent information refers to the proposition that [e]yewitness testimony about an event often reflects not only what [the eyewitness] actually saw but information they obtained later on.” “Confidence malleability refers to the proposition than [a]n eyewitness’s confidence can be influenced by factors that are unrelated to identification accuracy.” (People v LeGrand, 196 Misc 2d 179, 180-181 [2002] [internal quotation marks omitted].)